Split (1)

train · 100k rows

id stringlengths 3 9	source stringclasses 1 value	version stringclasses 1 value	text stringlengths 1.54k 298k	added stringdate 1993-11-25 05:05:38 2024-09-20 15:30:25	created stringdate 1-01-01 00:00:00 2024-07-31 00:00:00	metadata dict
238639249	pes2o/s2orc	v3-fos-license	Recognition of Fine-Grained Walking Patterns Using a Smartwatch with Deep Attentive Neural Networks Generally, people do various things while walking. For example, people frequently walk while looking at their smartphones. Sometimes we walk differently than usual; for example, when walking on ice or snow, we tend to waddle. Understanding walking patterns could provide users with contextual information tailored to the current situation. To formulate this as a machine-learning problem, we defined 18 different everyday walking styles. Noting that walking strategies significantly affect the spatiotemporal features of hand motions, e.g., the speed and intensity of the swinging arm, we propose a smartwatch-based wearable system that can recognize these predefined walking styles. We developed a wearable system, suitable for use with a commercial smartwatch, that can capture hand motions in the form of multivariate timeseries (MTS) signals. Then, we employed a set of machine learning algorithms, including feature-based and recent deep learning algorithms, to learn the MTS data in a supervised fashion. Experimental results demonstrated that, with recent deep learning algorithms, the proposed approach successfully recognized a variety of walking patterns, using the smartwatch measurements. We analyzed the results with recent attention-based recurrent neural networks to understand the relative contributions of the MTS signals in the classification process. Introduction As wearable devices are gaining popularity, wearable-based human activity recognition (HAR) has attracted increasing attention. Some fundamental functionalities have been adopted by many consumer smartwatches. For example, the device may encourage us to stand up if we sit still for a long time or request an SOS if we fall while alone. With the recent advances in sensors and wearable technologies, many studies have investigated using smartwatches as data-collection equipment [1][2][3][4]. To date, many HAR studies have focused on the coarse-grained classification of human movements, such as walking, running, sitting, and lying, each of which is a distinct activity. However, in various situations, it is often necessary to recognize fine-grained movements. In some cases, fine-grained classification would make computational experiences contextually aware [5]. For example, differentiating regular walking from inclined walking-walking on steps or on a uniform slope-may be required for the precise calculation of human energy expenditure [6]. In addition, recognition of a slight tremor when walking would make screening processes, such as for Parkinson's disease [7], more precise. In a similar context, we focus on the fine-grained classification of walking, which is a fundamental movement that comprises the largest proportion of humans' daily movements, and propose a system that can recognize predefined walking styles in a supervised manner. To that end, we defined 18 different walking styles, such as regular walking, carrying Many applications require fine-grained activity recognition; however, achieving a high recognition rate is challenging, because similar movements produce similar signals. A recent work reported that errors occurred when its system classified similar movements that involved walking patterns, e.g., differentiating regular walking from vacuum cleaning [2]. Weiss et al. [4] proposed a system that classifies various everyday activities using a consumer smartwatch. They reported that recognizing similar hand-oriented eating activities, such as eating pasta and soup, was challenging. Kwapisz et al. [8] also proposed a system to classify similar walking activities, including regular walking, and ascending and descending stairs. However, ascending and descending stairs were frequently evaluated as identical movements. Extensive feature-engineering work may mitigate such recognition issues; however, finding the ideal set of features for a classification process would be time-consuming [9,10]. Thus, classification with manually defined features may not be able to capture subtle differences in similar but different complex temporal patterns. To address the challenges in recognizing fine-grained activities, we adopted recent deep neural networks, such as onedimensional convolutional neural networks (Conv1D); gated recurrent neural networks (RNNs), such as long short-term memory (LSTM); and gated recurrent units (GRU). Although deep learning algorithms can learn complex and hierarchical features automatically from raw multivariate timeseries (MTS) data, the learning process is normally not designed to explain how its internal model works. To learn an interpretable representation and visualize the indicators of the raw data that seems influential in the model's evaluations, we further utilized attention-based neural networks. The primary contributions of this paper are as follows: 1. We defined a set of fine-grained walking styles that appear every day and proposed a wearable system that can recognize these predefined patterns in a supervised fashion. 2. We conducted an experiment to validate the feasibility of an intelligent wearable system with feature-based machine learning and recent deep learning algorithms, including attention-based deep neural networks. 3. We visualized and analyzed the parameters in the attention layer, which indicate the extent to which the classification result would depend on input signals from different time steps. Fine-Grained Recognition of Walking Activity Although quality of walking is used as a measure of the healthiness of a person [7,11,12], few studies have undertaken detailed classification of walking motion, as summarized in Table 2. In an earlier pioneering work, Bao and Intille [13] proposed a system that classifies daily movements, including activities related to walking, e.g., regular walking, walking while carrying items and ascending stairs, using the measurements from multiple on-body accelerometers. They found that overall recognition accuracy was highest when a decisiontree classifier was used for the task. They also envisioned that machine learning algorithms could be used to recognize different types of walking styles, such as walking slowly and walking briskly. In another study, a smart band-based wearable system was proposed to recognize five different walking styles, such as while texting or calling, with hands in pockets, whilst carrying a suitcase and regular walking, and achieved high and robust classification performance with a support vector machine (SVM)-based classification model [14]. Another previous work proposed a wearable system that utilized gait phase information [15]. Based on the walking distance-estimation algorithm and a decision-tree model, their system successfully recognized three different walking strategies, such as regular walking, walking upstairs and walking downstairs. Interestingly, another previous work demonstrated that acceleration information could be used to recognize differently inclined surfaces in a supervised fashion [16]. They proposed using customized time-frequency domain features to recognize different inclined walking based on a Gaussian-mixture-model classifier. Their experimental results demonstrated its remarkable classification accuracy. They also emphasized that the normalization process for features is crucial to minimize individual variation. A HAR system, based on a body-worn smartphone, was proposed in another recent study [17]. The proposed deep neural network learned the features successfully in an end-to-end fashion, after turning raw input signals into a multi-channel image using Fourier and wavelet transformations, resulting in high classification performance. Table 2 summaries previous studies on walking-related activity recognition. Smartwatch-Based Activity Recognition With the advances in sensor and wearable technologies, studies using smartwatches to recognize human activities have been increasing. In real-life situations, using a smartwatch to capture human activity is advantageous as compared to using a smartphone, in that a smartwatch it is normally placed on a specific body part (e.g., wrist) and does not interfere with body movements. One crucial but implicit assumption of using a smartwatch to recognize various human activities is that different types of activities would result in different hand movements; thus, types of the whole-body activities could be recognized (or observed) differently using measurements from the smartwatch. Based on this assumption, there have been numerous studies on HAR using recent smartwatches, particularly during the last decade. For example, an earlier study investigated the possibility of using a smartwatch to recognize 18 different everyday activities [20]. Remarkably, they achieved high accuracy and F m by proposing a stacked architecture, comprised of a convolution neural network (CNN) and LSTM. In another previous work, Mekruksavanich et al. [21] proposed a smartwatch-based system that can recognize six different human activities, i.e., sitting, standing, lying, walking, walking upstairs and walking downstairs, in the context of preventing office workers syndrome. With nine different selected features and an ensemble model, they achieved 93.5% classification accuracy. In a follow-up study, they used an LSTM-based deep neural network and achieved 96.3% classification accuracy [22]. A recent work explored and validated the feasibility of sensing hand-oriented activities using consumer smartwatches [5]. Based on an analysis of the spatiotemporal aspect of inertial hand movements using a recent deep CNN model, they achieved 95.2% accuracy across 25 fine-grained everyday hand activities. Although we have summarized relevant recent studies, it is important to note that research into smartwatch-based activity recognition systems is in an early stage. In this paper, we assumed that different types of walking activities generally involve different dynamic hand motions, as shown in Figure 1. Note that different walking strategies would result in different arm-swing patterns. From this perspective, we hypothesized that differences in MTS motion signals from different walking patterns could be learned by machine learning algorithms. To validate our hypothesis, we first developed an in-telligent wearable system that leverages recent advances in artificial neural networks. Then, we conducted an experiment in which participants were asked to walk as instructed with the device on their wrist. We will describe the experiment and the results in the following section. Experiment In this section, we first describe the wearable system developed for the proposed fine-grained activity recognition task. As described in the previous section, we focused on the wrist-worn smartwatch as walking patterns affect the hand motions while walking, differently from the previous studies that focused on the sensors attached to the leg [15] for recognizing the walking patterns. We then describe the activities defined in this study and the experimental procedure conducted to validate the proposed approach's feasibility. Equipment In the data collection process, we used a consumer smartwatch (DW9F1 by Fossil Group, Inc., Texas, USA) as the sensing device and a smartphone (Galaxy Note 20 by Samsung Electronics Co. Ltd., Korea) as the host device. For the smartwatch, we developed a custom application to capture the inertial movements of the hand in the form of MTS data using Wear OS by Google. Here, sensor values from the built-in motion sensors (e.g., triaxial accelerometer and gyroscope) were captured at every 20 ms. For the smartphone, we developed a custom host application to manage smartwatch application remotely over the Bluetooth low-energy (BLE) connection. With the host application, the experimenter can assign a label to the motion, take notes for the experiment, and control the start and end of the capture process remotely. Figure 2 shows the smartwatch device used in this study (left) and an example of the custom application running on the smartwatch (right). Activity Definition We defined a set of 18 different walking styles (Table 1 and Figure 3) that are used frequently in daily life. For motion classes C4 and C5, we asked the participants to read arbitrary content displayed on the smartphone while walking. For motion classes C6, C7, and C8, the participants walked while holding a 2-kg dumbbell (approximately 4.4 pounds) in the left, right, and both hands, respectively, to simulate holding a heavy load (e.g., groceries). Table 1. Class index is displayed at the bottom-right corner of each picture. Class #C17 (i.e., doing something while sitting and standing) was added as the reference class. Problem Definition Given the MTS input data x = x 1 , x 2 , . . . , x T ∈ R T×D , the machine learning systems for activity recognition attempt to estimate y ∈ R M , i.e., a type of activity from a predefined set of activities. Here, x t ∈ R D represents the t-th measurements, T and D (=6 in our case) represent the length of the signal and the dimension of the sensor data, respectively, and M denotes the number of activity types. Figure 4 shows the pipeline of the machine learning process used in this study. Figure 4. Pipeline of the machine learning process. A feature-based approach, in which machine learning step is preceded by a feature engineering process is used as the baseline models of deep learning-based approach. Data Collection Thirty-six subjects (20 to 62 years old; average age: 27.91; standard deviation: 11.57 years) participated in this experiment. Note that all participants self-reported being right-handed. In this experiment, the participants wore the smartwatch on their non-dominant hand (i.e., the left wrist). The participants were asked to walk according to the instructed walking styles. For class C0, we instructed participants to walk at a self-paced speed but not at higher intensities exceeding moderate levels. Most participants walked at least one lap around the 400-m campus track. For classes C15 and C16, the participants were moved to stairs, and for classes C13 and C14, the participants walked up and down ramps (inclined approximately 10 degrees), respectively, on the university campus. Although the experiment was conducted in different seasons (winter to summer), the amount of data obtained for classes C2 and C3 (walking with an umbrella) and class C1 (walking on thick snow) was relatively small compared to the other cases because specific weather conditions were required for data collection. In addition, a relatively small amount of data was collected for class C12 (jogging) because this task was performed in a shorter time over the same distance. Note that we instructed the participants to stop the trial whenever they felt uncomfortable, to avoid becoming tired after the experiment. The total time taken for each class is shown in Table 3. Cumulatively, we collected a total of 45.18 h (std: 0.72) of data from the 36 participants. Data Segmentation As described in Section 3.1, labelled information was assigned by the host device during the experiment. The collected MTS data were normalized by removing the mean and scaling to unit variance on each axis. The preprocessed data were then segmented using two different partitioning windows (T = 100 and 150 samples, accounting for 2 and 3 s of movement, respectively) without overlaps between adjacent segmentations. Here, we selected a motion segment length of T = 100 and 150 because common walking activities have a cycle of less than 2-3 s. Note that we did not align the signals according to the walking phase so that the machine learning models could learn features from each activity regardless of the activity phase ( Figure 5), a viable strategy according to a recent study [17]. Feature-Based Machine Learning Rather than relying on time-consuming feature-selection tasks, we employed the tsfresh library [9] to extract statistically significant timeseries features. The tsfresh library provides highly parallel feature selection algorithms based on the Benjamini-Yekutieli procedure [24], which is a false-discovery-rate-controlling procedure. In the feature-extraction process, a comprehensive number of features (e.g., 4686 = 781 × 6 features in our case) was extracted from the segmented MTS signal x ∈ R T×D . We then selected the 180 most-significant features. Here, approximately 30 features could be extracted for each axis based on the significance hypothesis test. The entire feature extraction process is illustrated in Figure 5, and Table 4 shows the 12 most significant features based on the results of the feature significance hypothesis test. As the baseline, we used a set of feature-based classifiers, including naïve Bayes (NB), support-vector-machine (SVM) [25], and random-forest (RF) [26] classifiers. The NB classifier is a probabilistic model based on Bayes' theorem [27]. The NB classifier is applicable to many practical problems; however, its performance often degrades due to the naïve assumption that features are conditionally independent and contribute equally to the output. The RF classifier utilizes ensemble learning, which is a machine-learning technique that combines many decision-tree classifiers. The RF classifier can handle highdimensional data efficiently and can mitigate the overfitting issue [28]. The SVM classifier is a machine-learning tool that is effective at classifying high-dimensional data [25]. In this study, the radial basis function (RBF) was used as the kernel function. Figure 5. Flow of the feature-extraction and -selection processes using the tsfresh library [9]. Final feature set is selected according to the p-values from the feature significant test. We adopted Conv1D, LSTM, and GRU to learn features and classify the segmented MTS signal x ∈ R T×D . In addition, we employed attention-based LSTM and GRUs to learn an interpretable representation that describes which parts of the input sequence receive the model's attention during classification. We adopted the attention mechanism, initially devised for machine translation tasks, for densely visualizing the machine attention to explain and interpret how the models come to a decision. Conv1D A convolutional neural network (ConvNet) is a particular kind type of artificial neural network comprised of multiple building blocks, e.g., alternating convolution and pooling layers to learn features, and fully-connected layers for classification and regression. A ConvNet extracts local features efficiently at a specific hidden layer by limiting the size of the receptive fields of filters (i.e., sparse connectivity). It also learns the spatial hierarchies of features using stacked deep-layer structures. Especially during the last few years, it has successfully demonstrated its capability to learn features from different types of information, such as regular image, spectral data [5,17,29], 3D volumes [30], etc. In a onedimensional convolutional neural network (Conv1D), convolutional kernels are convolved with the layer input over a single temporal/spatial dimension [31,32] to produce latent features. Conv1D can learn hierarchical features with low computational complexity, as the major operation is a simply weighted sum of two one-dimensional arrays [33], it is widely used in many practical sequence classification tasks, e.g., sentence classification [32], earthquake detection [34], surface recognition [35], context understanding [36], and realtime electrocardiogram monitoring [37]. Similar to a recent work [38], we set all the kernel sizes (i.e., the length of the 1D convolution window) as 3 and the stride length of the convolution as 1. LSTM The standard RNN with the traditional tanh unit suffers from the vanishing and exploding gradient problem, which makes difficult its learning long-term dependencies. LSTM was proposed to mitigate this issue. LSTM can learn long-term dependencies using memory-cell and gate units [39], and LSTM-based architectures have been employed in many sequence classification applications [35,36]. The memory cell stores information taken from the input and previous cells over the given period. This information is controlled by the gate units, i.e., update, forget, and output gates. GRU Similar to LSTM, the GRU [40] performs better than the basic RNN in many sequence transduction tasks, e.g., language modelling [41], torque generation [42], and many sequence classification tasks [36,43]. For the GRU-and LSTM-based architectures, we stacked recurrent cells two times (i.e., stacked two-layer GRU/LSTM [44]) to retain more long-term dependence information. The dimensionality of the output space of the recurrent hidden states was set to T, identical to the length of the input signal x ∈ R T×D . GRU and LSTM with Attention Mechanism Although gated RNNs, e.g., LSTM and GRU, and Conv1D have demonstrated their effectiveness in various sequence classification tasks, it remains difficult to explain and interpretate how the models come to a decision. Thus, for the proposed classification task, we utilized attention-based RNNs, which are typically applied to a variety of sequence transduction tasks in which alignments between different modalities must be learned [45][46][47]. Here, we adopted a multiplicative attention mechanism, which reduces encoder/decoder states to an attention score via a simple matrix multiplication [46]. As shown in Figure 6, our network comprises an LSTM/GRU-based sequence encoder, an attention layer, and a classification layer. Given the MTS input data x = x 1 , x 2 , . . . , x T , where x t ∈ R D represents the t-th measurement, the sequence encoder generates a sequence of hidden states a = a 1 , a 2 , . . . , a T , where a t ∈ R h represents the output of the t-th data point. The context vector, which is a weighted sum of a and captures relevant source-side information to predict the label of the input signal, is calculated by multiplying attention weights α with the encoder outputs a as follows. Here, α T,t describes the amount of attention thatŷ T should pay to the input feature at time t (i.e., a t ). As shown below, the alignment score is normalized with a softmax layer to produce the attention weights. Here, score(·) is a bilinear function, which compares the two hidden states, and W a is the trainable weight matrix of attention. The length of the alignment score α is T. Differing from attentional encoder-decoder problems [45,46], in our classification problem, a T is the last hidden state of the encoder network because our problem does not involve any decoder structure. A similar approach was used in recent studies [48,49]. The attentional hidden state h T is produced by concatenating the context vector c T and the last hidden state a T as follows: Then, the attentional vector h T is used to calculate the probability and label of the outputŷ T as follows. y T = argmax y p(y x) For the cost function of all deep learning-based approaches, we employed cross entropy between measured values, y, and estimated values,ŷ, which is defined as follows: where m is the batch size. Also, we added a dropout layer to the hidden layer output of all the deep networks to prevent overfitting. The Adam optimizer (with a learning rate of lr = 10 −3 , β1 = 0.9, β2 = 0.999) is used to train all of the deep learning-based models outlined to minimize cross-entropy loss [50]. Classification Results We use F1 score in the evaluation of each class, defined as harmonic average of precision (P) and recall (R), and weighted F1 score as the primary performance metric. To compute mean F1 score (F m ), we weight the per-class F1 scores by the number of instances for each class. Here, N c is the number of samples that belong to class c, and N tot is the total number of the samples from C different classes. Table 5 shows the classification accuracies and F m obtained from the experiments, and Table 6 shows the mean and standard deviation time required for inferencing a single data sample ∈ R T×D . Confusion matrices of the results from feature-based and deep-learning algorithms are shown in Figure 7. 782) respectively. There was no significant performance difference with the addition of attention. The benefits of an attention mechanism will be discussed in Section 5. Detailed classification performances are listed in Table 5. We also examined the high-dimensional internal features (D = 64 in our case) learned by our deep neural networks, such as Conv1D, LSTM, GRU, LSTM + Att, and GRU + Att, using t-distributed stochastic neighbor embedding (t-SNE) [51]. The two-dimensional embeddings projected from the last fully-connected layer are shown in Figure 8. Blind Test We collected an additional blind test dataset to further evaluate the robustness of the proposed system. The blind test data was collected from the subjects who did not participate in the experiment. We obtained the blind test dataset in a comparable but not identical environment to the training data because the blind test dataset was obtained assuming real-world conditions (e.g., flat walkway and field tracks on campus). Table 7 shows the total time spent on each class. Cumulatively, we collected a total of 35.90 min of data for the 18 classes, which is approximately 1.99 min (std: 0.87 min) for each class. The sampling rate was set to 50 Hz, the same as for the training dataset. Figure 9 shows the confusion matrix (left) and the corresponding t-SNE visualization of the blind test set using the Conv1D (upper) and LSTM (bottom) model as a classifier (right). Correctly classified data is marked with a filled circle and incorrectly classified data is marked with a cross. Classification Performance In general, based on the overall classification results, deep learning-based approaches successfully learned features from the different fine-grained walking styles defined in our study. During the test phase, it is noticeable that the LSTM/GRU-based approach demonstrated the highest accuracies and F m , i.e., greater than 96%, in both segmentation conditions. In our study, the addition of an attention layer did not significantly affect classification performance. Conv1D also exhibited high accuracies and F m over 96% when the length of the segmentation window was T = 150 (approx. 3 s). The most challenging activity to recognize was C8 (p: 88.027, r: 86.442) when LSTM was utilized. In contrast, feature-based approaches demonstrated lower classification performances over almost all the activities despite of the extensive feature-engineering process. Therefore, except for the SVM, it is apparent that the feature-based machine-learning models adopted in our study do not have sufficient capacity for learning the features from proposed finegrained motion dataset. Regarding the blind test described in Section 4.2., the accuracy (F m ) was significantly reduced by 9.686 (8.712) percent in the case of Conv1D compared to those from the test dataset. Although our approach validated the feasibility of the proposed learning scheme, robust recognition of some classes, such as C5, C8, C13, and C14, was found to be challenging as shown in Figure 9. More specifically, we found that C5 (walking phone right) was misclassified as C0 (regular walking) when the Conv1D model was used. This may be because there were differences in the degree to which participants focus on their smartphones, although they were asked to read the arbitrary contents displayed while walking. Also, we found that walking with a dumbbell in both hands (C8) was confused with walking with a dumbbell in the left-hand (C6). In addition, walking uphill (C13) was somehow confused with walking with a dumbbell in the right-hand (C7) and walking downhill (C14) was confused with walking downstairs (C16). An earlier work [8] reported a similar misclassification issue: ascending and descending stairs were frequently evaluated as identical movements. Noting that walking with something in the right hand and walking on inclined/stepped surfaces were successfully recognized in the training and test datasets but not in the blind test dataset, we plan to collect more data on these activities from diverse users to make our model more robust. Except for these classes, the rest of the classes' classification performance was better than or similar to the test dataset results. The blind test dataset, on the other hand, was analyzed using a modest amount of data. As a result, additional research with data from the various distributions is required. Attention Mechanism Learning an interpretable representation is crucial in many machine-learning tasks. A deep learning algorithm has an advantage of extracting features from the raw data; however, typically, understanding the relative contributions of the input data is a challenging task. To mitigate this issue, the concept of attention was introduced in earlier studies [45,46]. In this paper, we incorporate an attention mechanism, originally devised for neural-machine-translation tasks [46], into our classification model to learn an interpretable representation that describes which parts of the input data are receiving the model's attention. Different from recent studies on attention-based HAR systems [52][53][54], we further focus on densely visualizing and analyzing the attention weights along with the raw sensor input signal, x ∈ R T×D . Figures 10 and 11 are examples of visualization of attention vectors, α ∈ R T , highlighted in the bottom of each figure. The darker the highlighted bar, the more attention the attention vector received from the model during the inference phase. Note that attention values are formed in a continuous manner. In other words, a machine-learning model takes a collection of adjacent parts of input signals, rather than discrete parts of the signals, during the training and inference phases. This may be because input signals from specific intervals contribute to the calculation of the context vector, which captures relevant source-side information required to predict the label of the given MTS input signals. Figure 10. Example of MTS input signals from three different walking activities with temporally aligned attention vectors highlighted. The darker the highlighted bar, the more attention it received from the model, and thus contributing more during the inference phase. If the repetition cycle of the exercise was long, this example indicates that attention peaks shown in purple were formed at a slow cycle. Figure 11. Examples of input signals from walking with something in the right hand, with temporally aligned attention vectors highlighted. Walking with Something in the Right Hand Activities with something in the left-hand are relatively easy to recognize in that sensor values are recorded in the smartwatch worn in the left hand. There was little confusion reported between walking with an umbrella in the left and right hand (C2/C3) and between walking with a phone in the left and right hands (C4/C5). Although we initially assumed that it would be challenging to recognize cases in which the objects are being held in the right hand, it turned out that the proposed system could successfully recognize these activities, i.e., walking with an umbrella in the right hand (C3), walking with a phone in the right hand (C5), and walking with a dumbbell in the right hand (C7). This may be because our whole-body motion, including that of the left hand, is somehow affected by the constraints imposed on the right hand. For example, holding an umbrella or a heavy load in the either hand normally affects our dynamic walking patterns, such as spatiotemporal-stride and arm-swing parameters, significantly. Figure 11 shows examples of input signals from walking with something in the right hand (C3/C5/C7) with temporally aligned attention vectors highlighted. Note that the darker the highlighted bar, the more attention it received from the model; thus, contributing more during the inference phase. As shown in Figure 12, the two-dimensional feature embeddings from these activities (C3/C5/C7) are well clustered in distribution and separated those from other types of activities, including regular walking (C0). However, as we said in Section 5.1, our system is unable to detect all walking behaviors during the blind test. C8 (p: 100.0/86.667, r: 26.316/22.807 when Conv1D/LSTM were used) was, for example, mistaken with C6 (p: 46.213/29.605, r: 100.0/100.0 when Conv1D/LSTM were used). This could be because typical motion aspects (for example, swinging the left arm slowly due to a heavy load in the left hand) are invariant to right-hand motion. Figure 13 exhibits example input signals with attention weights aligned when our system misidentified C8 as C6 during the blind test. In contrast to Figures 10 and 11, which show examples of when the recognition process was correct, attention weights are not routinely and densely formed in Figure 13. Evaluation on Walking-Related Datasets We compared the classification results to those of other publicly available datasets. First, we used the PAMAP2 dataset (Physical Activity Monitoring for Aging People 2) [55], which includes 12 daily physical activities measured by on-body sensors attached to three different body parts, the hand, chest, and ankle. This dataset, interestingly, contains walkingrelated activities, such as walking, running, Nordic walking, ascending/descending stairs, and vacuum-cleaning. To achieve a temporal resolution comparable to our dataset, we downsampled the PAMAP2 dataset from 100Hz to 50Hz. The data was segmented into 3 s fixed-width sliding windows with no overlap. We also created a hand-oriented subset (PAMAP2-hand) using measurements from a sensor attached to the hand. Second, we used the SBHAR dataset (Smartphone-Based HAR dataset with Postural Transitions), which is a multivariate time series data from 30 participants ranging in age from 19 to 48 years [56]. This dataset includes six basic activities (walking, walking upstairs, walking downstairs, sitting, standing, and lying) and six postural transitions (standing-to-sitting, sitting-to-standing, sitting-to-lying, lying-to-sitting, standing-to-lying, lying-to-standing). A smartphone mounted on the participant's waist served as an inertial motion-capture device, equipped with a triaxial accelerometer and a gyroscope operating at 50 Hz. For testing our approach with the SBHAR dataset, we segmented the measurements using a sliding window of 3 s, with 50% overlap. Third, we used the Daphnet freezing of gait (DG) dataset [7], which consists of inertial measurements (i.e., acceleration) from 10 Parkinson's disease (PD) patients who are experiencing freezing of gait (FoG), which manifests as a sudden and temporary inability to move. The DG dataset is collected while PD patients are walking using on-body sensors attached to 3 different body parts (ankle, knee, and trunk). We validated our approach by downsampling our DG dataset from 66Hz to 50Hz and segmenting it with a sliding window of 3 s without overlap. Table 8 contains detailed information used for the evaluation. For more information on each dataset, see previous studies [10,57], which extensively summarizes the public dataset. Table 8 shows performance in terms of weighted F1 scores (i.e., F m ) from the different public datasets along with ours. As shown below, we demonstrate that it is feasible to learn features from the walking-related activities, each of which is inherently bound to have similar temporal features, using the recent deep learning-based approaches. Although there is no significant performance improvements with the addition of attention, it enhances the explainability of the classification process. Limitations The proposed model demonstrated high accuracies and F m in recognizing activities on the test set. However, as discussed in Section 4.2, it is not guaranteed that similar recognition performance can be achieved in real-life scenarios because our data was collected in a controlled environment and from a limited number of participants. In fact, Bao and Intille [13] emphasized the importance of unsupervised and naturally collected data. They collected two different types of data. One type was collected in a semi-naturalistic environment, wherein the participants were asked to complete descriptive tasks. This setting allowed participants to move on their own to some extent. The other type was collected in a laboratory setting where the participants were instructed to execute several predefined activities. Since our data was also collected in a controlled setting, we plan to design experiments in which participants are allowed to move more naturally. Applications Fine-grained classification of walking styles would open a new venue for promising applications in diverse fields, such as providing contextual information tailored to a user's current situation, measuring precise energy expenditures during exercise, and monitoring abnormal activities. Assistance for Distracted Walkers As smartphones become more common, people often look at their smartphone screens, even when walking. Consequently, a distracted walker may get into an accident. A part of our study, i.e., differentiating walking activities while looking at the smartphone screen (C4/C5) from regular walking (C0), can be utilized to help walkers. For example, wearable assistants based on our approach could provide distracted walkers with warnings when they enter a busy street. Identifying or recognizing cognitive loads while walking using wearable devices would be interesting future work. Contextual Applications Furthermore, recognition of the availability of the user's hands, e.g., walking with dumbbells in both hands (C8) and walking with an umbrella in either hand (C2/C3), would be useful for those who cannot use their hands to manipulate smart devices. For example, wearable applications could read incoming messages or open car doors automatically if the system recognized that a user was moving with luggage in both hands. Encouraging Fitness Recognition of fundamental activities, such as running and walking, are already embedded in modern consumer smartwatches. For example, the device may encourage us to stand up if we sit still for a long time, and the device can recognize whether we are walking or running for fitness. As we investigated throughout this study, our activities could be recognized in much finer detail. A wearable system may encourage users who are exercising to walk faster if they are walking slowly or with their hands in their pockets. Without loss of generality, the proposed approach can be extended to summarize the recorded activities into a set of fine-grained activities, enabling personalized fitness suggestions and encouragements. Conclusions Assuming that hand motions are an important part of human walking activities and thus have different spatiotemporal characteristics according to the walking styles, we propose a wearable system that can recognize fine-grained walking patterns. To that end, we defined 18 different everyday walking styles and developed a wearable system that can capture a user's body motions from their hand motion in the form of MTS signals. Then, we employed a set of machine-learning algorithms, including feature-based algorithms and recent deep-learning algorithms to learn the MTS data with the predefined walking patterns in a supervised fashion. With our model, the LSTM-based approach demonstrated the best classification results in terms of accuracy (F m ) of 97.158 (97.156). However, deep-learning-based approaches, including Conv1D, LSTM, GRU, LSTM + Att, and GRU + Att, generally exhibited higher classification performance, i.e., accuracy and F m greater than 95%. Despite of our extensive feature engineering work, feature-based approaches demonstrated poor classification performances overall. One remarkable finding from the experimental results was that walking activities with something in the dominant hand can be recognized even when the smartwatch is worn on the non-dominant side. Regarding the blind test, the classification results of accuracy (F m ) were 87.290 (88.259) when Conv1D was employed. Our model has trouble robustly recognizing specific walking patterns, such as walking with something in the right hand and walking on inclined/stepped surfaces, according to the findings of the blind test. To resolve this generalization issue, we plan to collect more data on these activities from diverse users to make our pretrained model more robust. Using the attention-based neural networks, we further analyzed the classification results to understand the relative contributions of the MTS signals used in the classification process. In the application section, we explored a set of wearable applications that utilize the proposed fine-grained walking activity-recognition scheme. Future studies will focus on increasing the robustness of the model and extending the proposed approach to diverse healthcare applications. Informed Consent Statement: Written informed consent has been obtained from participants in the study. Data Availability Statement: We cited the details of each dataset in the document. Acknowledgments: The authors would like to thank the reviewers for all of their constructive and insightful comments in relation to this work. Conflicts of Interest: The authors declare no conflict of interest.	2021-10-13T13:22:39.239Z	2021-09-24T00:00:00.000	{ "year": 2021, "sha1": "b8747fbab52a1f14bc6a7db2855dd560ec48d3bd", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1424-8220/21/19/6393/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "d295775530faae7030ac8bf0598f1cc0e505d692", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Computer Science", "Medicine" ] }
248437461	pes2o/s2orc	v3-fos-license	STRATEGIES AND PRAGMATIC EQUIVALENCE OF THE WORD FUCK TRANSLATION IN THE WOLF OF WALL STREET MOVIE The choice of strategy for translating the word fuck in The Wolf of Wall Street movie has its own impact on the equivalence of pragmatic meanings in the target text. This study aims to explain: (1) form, pragmatic function, and translation strategy of the word fuck , and (2) the implication of the word fuck translation strategy on pragmatic equivalence in The Wolf of Wall Street movie. The words of fuck were identified from the both English and Indonesian subtitles. Then, the data were categorized and the implications were investigated. The results indicate that there are 506 words of fuck which are classified into three forms, namely: interjection, emphasizer, and expletive slot filler. The functions conveyed by these taboo words are: catharsis, abusive, and social. These taboo words were translated using omission (72.9%), cultural substitution (15.0%), neutral words (11.5%) and loan word (0.6%). The omission and the neutral word strategy largely eliminated the function of the word fuck , whereas the loan word and cultural substitution strategy were able to maintain the function of the word fuck . However, the very large percentage of use of the omission strategy makes the translation results lose much of its pragmatic function and emotive value. Introduction The word fuck is a taboo word that is popularly used by English language speakers. Etymologically this word comes from Latin futuere (or pungere or battuere), French foutre, German ficken which all have the contextual meaning of 'hitting' or 'beating' and the meaning of 'sexual activity' (Wajnryb, 2005: 52). It also has links to widely spoken German words (Middle Dutch fokken, Norwegian fukka, and Swiss focka), all of which mean 'to hit' or 'push ' (p. 53). In An Encyclopedia of Swearing, Hughes (2006: 188-189) notes that the word fuck firstly appeared not among the lower classes but among Scottish poets and aristocrats. William Dunbar is recorded as the first person to use it in his poet, namely in 1503, when he said "He wald have fukkit". Sir David Lindsay, a Scottish satirist, also used it in 1535 when criticizing the hypocrisy of the pastor, "Bischops ... may fuck their fill and be vnmarryit" in Satire of the Three Estates. Since 1857, the word has been categorized as obscene libel or 'obscene slander'. This means that a publisher can be sued if it publishes a book containing that word. Based on the classification of Ljung (2011: 5) the word fuck can be divided into several forms, including: interjection or interjection, such as fuck! and motherfucker!; emphasizer or to emphasize especially question sentences, for example what the fuck is that?; and expletive slot filler or in the form of filler which actually has no meaning at all, for example He is not my fucking friend. In any form of use, this taboo word is strictly prohibited, especially in official contexts and situations. State and religious institutions usually play a role in this prohibition because the word fuck is a lexicon that contains offensive emotions (Jay, 2009: 153). However, the measure of taboo discrepancy depends on contextual and pragmatic variables, such as: the interpersonal relationship background of the speaker and speech partners, social and physical background, and choice of diction and tone of speech (p. 154). The word fuck, which was originally so avoided, is now increasingly appearing in world films, especially Hollywood. Hughes (2006: 192-193) states that this word has even been accepted as part of popular culture, especially in movies and television shows. Some movies that started the use of the word fuck included: Apocalypse Now! (1980), Platoon (1987), andFull Metal Jacket (1988). Guinness World Records records The Wolf of Wall Street as the movie which contains the most of taboo words (Park, 2018). This movie tells the life story of a stockbroker named Jordan Belfort. The word taboo in this film is not spoken in the suburbs but on Wall Street, New York. The abundant mount of taboo words in this film portrays another side of the world of investing which is glamorous yet full of deception. Translating the word fuck in a film is certainly a challenge. However, the translator needs to consider the norms and culture of the target language community. Unlike Hollywood, Indonesian cinema avoids using taboo words. Foreign movies which contain taboo words will be adapted to the norms and culture prevailing here through a censorship process. Therefore, translators often have a tendency to soften taboo words found in foreign films compared, for example, by translating them using taboo words in the target language with the aim that the subtitles can be well received by the public in the target language. The obstacle that arises later is that this softening effort often has a negative impact on the translation results, including the loss of the emotive value. The function of the presence of taboo words in the source text will then disappear in the target text. This clearly interferes with the pragmatic equivalence of the translation. The relationship between the translation strategy and the shift in the function of this taboo word has not been noticed by many studies. Therefore, this study aims to fill that gap by explaining: (1) the form, pragmatic function, and strategy of translating the word fuck in The Wolf of Wall Street, and (2) the implications of the translation strategy on the function of the word fuck in The Wolf of Wall Sreet. Translation Strategies Understanding the function of the word fuck is an important thing in the translation process so that the translator can use the suitable translation strategy to achieve translation equivalence. Baker (2018: 24-45) summarizes the existing translation strategies and reformulates them into eight translation strategies: (1) Strategy using superordinate or general words. General words are very often to use when translating un-equivalent words in the target language. (2) Neutral or less expressive word strategy. This strategy is commonly used to avoid offensive words to soften the effects. (3) Cultural substitution strategy. This strategy is used for translating specific words that only exist in the culture of the target language in order to bring a similar impact with the word in the source language, even though they sometimes have different propositional meanings. (4) Loan word strategy with or without explanation. This strategy is usually carried out when the target language really does not have an equivalent word or the translator deliberately wants to introduce the meaning of the source text completely. (5) Paraphrasing using related words. Concepts in the source language are lexicalized or restated differently in the target language. (6) Paraphrasing using unrelated words. It is basically paraphrasing using by using unrelated words and concepts. (7) Translation strategy by means of omission. Omission will be possible if the word does not interfere with the core meaning of the clause or sentence. (8) Translation strategy using illustrations. It is the only translation strategy that uses non-linguistic elements, namely using visual aids. Pragmatic Equivalent in Translation The equivalence, according to Nida (1964: 12), is a process of searching for the closest natural words of the source language message, in terms of meaning and style. Meanwhile, according to Venuti (2012: 5) equivalence can mean truth, accuracy, adequacy and connectivity between meanings in source language and target language. Translation equivalence can be achieved by selecting the correct translation orientation. Both also introduced the types of translation based on their orientation. The ideology of translation foreignization and domestication introduced by Venuti (2008: 15) intersects with the equivalence of formal and dynamic translation from Nida (1964: 12). Foreignization and formal equivalence are oriented towards the source language by prioritizing the original meaning of the source text, while domestication and dynamic equivalence are oriented towards the target language by prioritizing the meaning and response of the audience to the target language. Pragmatic equivalence relates to the entanglement of meanings beyond the textual meaning and more to the contextual meaning. The contextual meaning of the word taboo is closely related to the function or reason for what the word taboo is used. It is these taboo-word functions that translators should maintain in the target language in order to achieve pragmatic equivalence. The functions of taboo words according to Wajnryb (2005: 30-40) include: (1) The catharsis function, which is the most instinctive or spontaneous function among other functions. A swear word that functions as catharsis usually comes out of someone's mouth when the person's feet accidentally trip over the table. This swear word expresses both the irritation and the pain that is being felt at that moment. (2) Abusive function, namely a function to verbally attack people. The abusive function has significant differences with the catharsis function in terms of the involvement of others. The swear word catharsis requires someone else as the object of the curse while the swear word catharsis does not. An example is the taboo word someone uses when he sees a reckless motorist on the road. The taboo word that is spoken is intended to curse the driver who could endanger him and others around him. (3) Social function, namely a function to show friendship. Taboo words that appear in an interaction aimed at other people but are not intended to be offensive can be an indication that speakers and the interlocutors have a close interpersonal relationship or are in a relaxed situation. Translation of Taboo Words in The Wolf of Wall Street Movie The Wolf of Wall Street is a movie listed in the Guinness World Records as the film with the most of taboo words (Park, 2018). The availability of abundant taboo data in this film makes this film used as an object of research in various subtitles, namely Park (2018) in Korean subtitle, Abu Sirrieh et al. (2019) and Hawel (2019) in Arabic subtitle, and Bram and Putra (2019) and Sari et al. (2016) in Indonesian subtile. Apparently none of the previous research concerned on how taboo translation relates to its pragmatic functions. Hence, this study aims at filling the gap by exploring that the choice of translation strategies has something to do not only with meaning but also with pragmatic equivalence. Method The formal object of this research is the word taboo fuck and its combination of words found in the material object in the form of the English and Indonesian subtitles of The Wolf of Wall Street movie. The movie, which was inspired by a true story, was made in 2013 and directed by Martin Scorsese. This film has been named the film with the most of taboo words by Guinness World Records (Park, 2019). Therefore, this movie is rated R or Restricted, which means when watching this film, children under the age of 17 must be accompanied by parents or adults. Data were collected by firstly watching the movie, then identifying the word fuck in the both subtitles, and encoding them. The collected words of fuck were then analyzed by categorizing word fuck based on its form (Ljung, 2011), pragmatic function (Wajnryb, 2005), and translation strategy (Baker, 2018); and linking the translation strategies used with its pragmatic function, whether they have implications on the pragmatic equivalent of the target text. The Form of The Word Fuck It has been found 506 fuck words and their variations in the The Wolf of Wall Street movie. These words come in a variety of forms. Using classification of taboo word by Ljung (2011: 5), the form of fuck in this movie could be categorized into: interjection, emphasizer, and expletive slot filler. The examples of interjection form are: fuck, fuckety-fuck-fuck, motherfucker, and fuck off as shown in the following utterances. The word fuck also appears in the form of a question that uses whwords. Here, this taboo word emphasizes the question sentence. An example of this emphasizer is shown below. ST: Where the fuck is it? TT: Di manakah itu? In addition, the word fuck appears in the form of expletive slot fillers or as filler words that actually have no meaning but can express the emotions of the utterance. Words which appear in this form are: fucking, and motherfucking. In utterances (6) and (7) these two words are not even translated into the target language at all, but this does not interfere with the unity of the meaning of each of these sentences. Omission Strategy The omission strategy was used 369 times or 72.9% and made it the most widely used strategy in translating the word fuck in this film. This strategy is mostly used in translating emphasizer and expletive slot fillers such as fuck in utterance (5) and fucking in utterance (6). Although the propositional meaning of a sentence is not reduced at all, the omission strategy eliminates the emotive meaning of these taboo words. Jordan Belfort's anger at the loss of his money and his lover's jewelery was on utterance (5) and Stratton Oakmont employee's frustration at utterance (6) towards FBI officers who caught him is not expressed in the target language subtitles. The greater loss of emotive value occurs in (8) where two taboo words are simultaneously removed from the translation. The taboo words cooksucking and motherfucking which are used together as adjectives which describe the same object, namely Steve Madden, shows the great emotion of Jordan Belfort to the person who deceived him in the sale of shares. ST: Even though I owned 85% of Steve-cocksucking-motherfucking- Madden Shoes TT: WaIaupun aku memiIiki 85% dari sepatu Steve Madden Baker (2019: 43) views that omission as commonplace in translation and in certain contexts is not detrimental at all. But on the other hand, the omission strategy or Molina and Hurtado Albir (2002) called this reduction in fact eliminating the emotive value of a sentence that the speaker intends to convey through intensive, so that it has the potential to produce inaccurate translation products (Sari et al., 2016: 97). Cultural Substitution Strategy The cultural substitution strategy is carried out by replacing the word BSu with a word that does not have the same propositional meaning but can have the same impact on the target reader (Baker, 2019: 30). This cultural substitution strategy is similar to Davoodi's (2009) taboo for taboo translation strategy because they both focus on producing an impact for the audience. As many as 15.0% of the word taboo fuck in the subtitle of the film The Wolf of Wall Street was translated using a cultural substitution strategy. This strategy translates the word fuck into a taboo word in the target language even though the literal meaning is so far away. The word fuck in utterance (9) is translated as affair which comes from the root word Satan which according to the Kamus Besar Bahasa Indonesia (KBBI) V means "the word to express the anger" (2019). Another form of cultural substitution exists in (10) where fuck can be translated into bad luck. Based on KBBI V (2019), bad luck falls into the category of harsh words and means "the wretched one; hell." Both fuck and bad luck have a literal meaning that is far removed from the literal meaning of fuck which according to the Cambridge Advanced Learner's Dictionary means "an act of having sex" (2008). Although most of the word fucking becomes expletive slot filler, some have a core meaning and are translated using cultural substitution in the form of a taboo word in the target language. The cultural substitution strategy is able to present a translation equivalence that is closer to the taboo word of the source language because this strategy prioritizes the impact felt by the audience from the translation results, not merely on the similarity of literal meanings. Neutral Word Strategy This strategy is used 58 times or 11.5% and is the third most used strategy in translating the word fuck variant in the film The Wolf of Wall Street. The verb phrase give a fuck has only one possible equivalent in Indonesian, namely peduli. The word care clearly doesn't have an emotive meaning or is not as expressive as give a fuck. In (12) there is a clear difference in the way of expressing meanings between English and Indonesian. In English, insults are not only conveyed in the form of interjection, nouns and adjectives, but also in the form of verbs. Whereas in Indonesian, taboo word is more in the form of interjection and nouns. ST: Nobody gives a fuck. TT: Tak ada yang peduIi. The word fucking in (13) is not in the form of expletive slot filler, it has the role of an adjective which gives information to the adjective in front of it which is convenient. Fucking here is translated as a word that has no equal emotive value, namely sangat. Strategy used in (12) and (13) can also be considered as euphemism as it softens the taboo words of the source text (Surya, 2014). TT: Bukankah sungguh nyaman bagimu? The choice of strategy with the use of neutral words is mostly driven by differences in the concept of lexicalization and expressive meanings in the two languages which are actually part of several general obstacles in the mismatches conveyed by Baker (2018: 19-24). This difference in literal meaning according to Fernández Dobao (2006) is not so important in translating taboo words because what needs to be paid close attention is the emotional equivalence and behavior of the speaker so that the impact felt by the audience on the translation results with the original text can be the same. Loan Word Strategy Only three times this strategy was used. The only word translated using this strategy is fucksville. This word has a very close connection with the situation and culture of the United States of America where many city names ended by -ville. Fucksville itself never existed in the United States. This word was made only to show Donnie's anger and ridicule towards Brad who acted as if he were a king. Therefore, Donnie calls him The Emperor of Fucksville. ST: Oh, my gosh. The Emperor of Fucksville came down from Fucksville to give me a pass! TT: Kaisar FucksviIIe datang dari FucksviIIe untuk beri aku ijin Iewat. There is no exact equivalent of fucksville in Indonesian. This is due to differences in regional naming systems in the United States and Indonesia as well as the flexibility of the word fuck itself. Indonesian does not have swear words, it can be positioned as any class of words such as the word fuck. Therefore, the translator uses this loan word strategy as an option because this strategy is appropriate to use when meeting words that have specific cultural meanings (Baker, 2019: 34). Yamayanti et al. (2016) suggest on their study that generally strategies used by translator are literal and idiomatic. Loan word is an alternative option when two options cannot provide the equivalence and accommodate the intention of the translator to maintain the cultural meaning of the word. Table 2 shows that the word fuck has a social function of 75%. Abusive function is only 21% and is a function that only appears at certain times unlike the social function that appears in almost all occasions. The least function used was catharsis as much as 4%. These taboo words pragmatic functions proposed by Wajnryb (2005) inform how each taboo word contain not only meaning bt also specific emotion. Table below shows that most of the word fuck and its combination of words are mostly used just not to offend or attack the interlocutor verbally, but to show that the taboo word, especially fuck, has become the norm and the character's speaking style. It can also prove that this film wants to show the dark and dirty side of Wall Street and the world of US stock investing in general. Strategy for the Omission of Social, Catharsis and Abusive Functions The choice of translation strategy used has implications for the pragmatic function of the word fuck, which in fact has undergone several shifts. All the pragmatic functions of taboo words are lost when the omission translation strategy is used. The fuck on utterance (8) Likewise, what happens with social functions as in speech (7). This utterance (7) was uttered by one of Stratton Oakmont's employees to his boss, Jordan Belfort, who was about to separate. The word fucking here shows how much the employee loves him. Yet this omission strategy keeps such emotions from conveying. The utterance in the target text becomes normal. While the impact of this strategy on catharsis function is shown in utterance (15). In this utterance, it can be seen that the surprise felt by the speaker is stronger in the source language with the existence of fucking as an expletive slot filler when compared to the target language. ST: Jesus fucking Christ! TT: Astaga! The high number of omission in this translation is understandable. According to Afandi and Cholsy (2018) in subtitle translation, omission can be inevitable due to: word limitation, the existence of audio and visual that can represent context, and cultural difference. Besides, according to Santaemilia (2008) translators do not carry out censorship without consideration. The translator usually has considered three things: the ideology of the translation, aesthetics and cultural aspects. The translator, in the context of this research, ignores emotive values to achieve other goals, for example in order to conform to the norms and culture of the community in the target language so that the translated subtitles can be well received. Neutral Word Strategy on Abusive and Social Functions A similar effect is caused by a translation strategy using neutral words. Like the omission strategy, this strategy also has a significant impact on the abusive function. In utterance (16) the translator only shows the main purpose of the utterance which is to expel people using the word go. Whereas the phrase fuck you to expel people is a very strong word and has a very strong emotive value. The social function of the word taboo in utterance (17) is lost because the word fuck is only translated according to its meaning without including its emotive value. Workplace situations that are full of taboo words yet warm and friendly are lost in the target text because they are translated using inexpressive words. ST: Let's fuck! TT: Ayo mulai! This result is in line with research by Abu Sirrieh et al. (2019) suggesting that euphemism, including the use of neutral word, and omission are more likely to use when translating text relating to sexual, and religion. This is to prevent the abusive function to transfer to target text. Cultural Substitution Word Strategy on Abusive Functions Cultural substitution and loan word strategies have a different impact. This cultural substitution strategy is the strategy most capable of maintaining the pragmatic function of taboo words because this strategy translates taboo words in the source language into taboo words in the target language. Therefore, all fuck words translated using this strategy actually perform abusive function. Motherfucker in (18) (2019) indicating that the most used strategy to translate taboo words in an English book to Indonesian language is paraphrasing using related words. This strategy can retain the emotion of taboo words in target text. Loan Word Strategy for Abusive Function Loan word strategy retains its full emotive meaning because this strategy does not involve any change as seen in utterance (14). This strategy is usually used to translate words with specific cultural meaning (Baker, 2019). However, this strategy actually has the potential to reduce the pragmatic function if the audience does not understand the loan words used. Audiences who do not understand the meaning of the word fucksville in utterance (14) will, of course, be confused so that they do not feel the abusive function of the word. The shift of these pragmatic functions greatly affects the equivalence of the results of the translation itself. The audience response when reading English subtitles and Indonesian subtitles is likely to be different. Yet it is this similarity of emotions and responses that should be pursued. Conclusion The word fuck in The Wolf of Wall Street appears in the form of interjection, emphasizer, and expletive slot filler. This taboo word occupies the following functions: social, abusive, and catharsis. The word fuck in this film is then translated using several strategies including: omission, neutral words, cultural substitution, and loan word. The excessive use of taboo words in order to show the dark and harsh side of Wall Street, especially Jordan Belfort's personal life, becomes invisible due to the many use of omission strategies. The choice of translation strategy is driven by differences in the linguistic aspects of the two languages, namely differences in the concept of lexicalization and differences in expressive meanings. In addition, the translator's intention to refine the subtitles to match the community norms in the target language has become one of the reasons. The translator's efforts to soften taboo words in Indonesian subtitles have an impact on the results of the translation itself. The large percentage of use of the omission strategy results in drastic loss of the pragmatic functions of the word taboo fuck. The function that has the most significant impact is the social function because this function is found the most in the movie. In addition, the abusive function also has a very pronounced impact. Emotive value that shows the speaker's anger is missing in the target text. Explosive emotions do not convey well in the target text. The result is that the utterance containing curses becomes just like ordinary speech.	2022-04-29T15:44:29.039Z	2021-11-19T00:00:00.000	{ "year": 2021, "sha1": "cd4788e795351e84a7d7bc76ff70258bb2ab200d", "oa_license": "CCBYSA", "oa_url": "https://jurnal.ustjogja.ac.id/index.php/JELLT/article/download/11422/4554", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "ae05c7f25868eb22ed0dd7a515845d850a1d3e9a", "s2fieldsofstudy": [ "Education" ], "extfieldsofstudy": [] }
270255439	pes2o/s2orc	v3-fos-license	Long-term outcomes and prognosis of neuroendocrine neoplasms of the head and neck: a cohort from a single institution Background Neuroendocrine neoplasm is a rare cancer of head and neck. This study aimed to evaluate clinical features, treatment outcomes, and prognostic factors of neuroendocrine neoplasm of head and neck treated at a single institution. Methods Between Nov 2000 and Nov 2021, ninety-three patients diagnosed with neuroendocrine neoplasms of head and neck treated at our institution were reviewed retrospectively. The initial treatments included chemotherapy (induction, adjuvant, or concurrent) combined with radiotherapy in 40 patients (C + RT group), surgery followed by post-operative RT in 34 (S + RT group), and surgery plus salvage therapy in 19 patients (S + Sa group). Results The median follow-up time was 64.5 months. 5-year overall survival rate (OS), progression-free survival rate (PFS), loco-regional relapse-free survival free rate (LRRFS) and distant metastasis-free survival rate (DMFS) were 64.5%, 51.6%, 66.6%, and 62.1%, respectively. For stage I–II, the 5-year LRRFS for patients’ treatment regimen with or without radiotherapy (C + RT and S + RT groups versus S + Sa group) was 75.0% versus 12.7% (p = 0.015) while for stage III–IV, the 5-year LRRFS was 77.8% versus 50.0% (p = 0.006). The 5-year DMFS values for patients with or without systemic therapy (C + RT group versus S + RT or S + Sa) were 71.2% and 51.5% (p = 0.075). 44 patients (47.3%) experienced treatment failure and distant metastasis was the main failure pattern. Conclusions Radiotherapy improved local–regional control and played an important role in the management of HNNENs. The optimal treatment regimen for HNNENs remains the combination of local and systemic treatments. Introduction Neuroendocrine neoplasm is a rare and heterogeneous tumor type (Rindi et al. 2022), comprising about 2% of all malignancies, with the most common site of lung (Oronsky et al. 2017;Rekhtman 2022).Primary neuroendocrine neoplasms of the head and neck (HNNENs) are rare, accounting for only 0.3% of all head and neck cancers (Ohmoto et al. 2021;Rindi et al. 2022).Due to the low incidence, studies describing features and management of HNNEN remain limited. The World Health Organization (WHO) 2022 Classification of head and neck tumors divides HNNENs into well-differentiated neuroendocrine tumors of head and neck (HNNETs) and poor-differentiated neuroendocrine carcinomas of the head and neck (HNNECs) (Mete and Wenig 2022).HNNETs are generally graded as G1, G2, and G3 based on proliferation, and HNNECs are sub-classified as small or large cell types according to cell morphology (Mete and Wenig 2022).Moreover, the widely accepted classification, which divided HNNETs into typical carcinoids and atypical carcinoids, is previously used. Treatment modalities, including surgery, radiotherapy, chemotherapy, or the combination of various treatments, have been used for patients with different conditions (Matsuyama et al. 2022;Mitchell et al. 2021;Strojan et al. 2019).However, there is no consensus on the optimal treatment strategy for HNNENs.Since the occurrence of HNNENs is sporadic, it is hard to perform large clinical trials.Thus, we performed this retrospective study with large sample size, aiming to assess the long-term outcomes and prognosis and further explore the role of radiotherapy in the treatment of HNNENs. Patient data A retrospective review of all patients with diagnoses of HNNENs at our institution between November 2000 and November 2021 was performed.This study was approved by the local ethics committee.A total of 93 patients with pathologically proven HNNENs were included in this study.All diagnoses were confirmed by two independent pathologists in our hospital.The TNM stage classification was evaluated according to the Staging Manual of American Joint Committee on Cancer (AJCC) or Union for International Cancer Control (UICC) for different location and determined by clinical and imaging examinations, including CT or MRI. Treatment Among 93 patients, 40 patients underwent chemotherapy and radical radiotherapy (C + RT group: 33 of them underwent induction chemotherapy, all 40 had concurrent chemotherapy, 10 of them received adjuvant chemotherapy), 34 patients had radical surgery combined with post-operative radiotherapy (S + RT group: 14 of them also received concurrent chemotherapy) and 19 patients underwent primary surgery alone followed with salvage treatment (S + Sa group). All patient treatment strategies were recommended by our multidisciplinary board.The recommended treatment for early disease was surgery alone or concurrent chemoradiotherapy.For those at high risk of recurrence after surgery, including those with positive or close surgical margins, or pathologically positive lymph node, postoperative RT was suggested and delivered at 4-6 weeks after the surgery.As for those with unresectable disease or large tumor burden by imaging examinations, induction chemotherapy followed with radical radiotherapy was recommended.The radical radiation dose of gross tumor volume (GTV) was about 70 Gy, and the clinical target volume (CTV) about 60 Gy.The median CTV dose of post-operative RT was 60.06 Gy, ranging from 45 to 64 Gy. The regimen for chemotherapy was mostly cisplatin-based combination regimen, typically etoposide plus cisplatin (EP). Statistical analyses All events (including death, progression, loco-regional failure, and distant metastasis) were measured from the date of diagnosis until clearly documented first-time failure, corresponding to overall survival rate (OS), progressionfree survival rate (PFS), loco-regional relapse-free survival free rate (LRRFS), and distant metastasis-free survival rate (DMFS).Categorical data were compared using the chi-square test.The survival data were estimated using Kaplan-Meier method.All survivals were compared using the log-rank test.Multivariate analysis using the Cox proportional hazard model was performed to identify the prognostic factors.All reported p values were two-sided, and p values below 0.05 were considered to be significant. Patient characteristics Among the 93 patients, there were 71 male and 22 female, with a median age of 53 years (range from 18 to 82 years).The most common primary sites were as follows: nasal cavity (25 patients), larynx (23 patients), paranasal sinus (19 patients), and nasopharynx (9 patients).The other sites included oropharynx, oral cavity, saliva gland, thyroid, and unknown original site.According to T stage, patients were divided in two categories: 33 patients were T1-2, 59 were T3-4 and one patient had unknown T stage.As for lymph node status, 38 patients had positive lymph node metastasis, and 55 had negative status (N0).The TNM stage distribution was as follows: there were stage I-II 26, stage III-IV 66, and unknown 1. Thirteen patients were classified as NET and 80 were NEC (Table 1).For S + RT group, 25 of the 34 patients underwent microscopically margin-negative resection (R0) and the others suffered positive surgical margins.For the 19 patients in S + Sa group, 18 received primary R0 resection.After the first-time recurrence, 6 received salvage surgery, 2 underwent salvage surgery plus post-operative radiotherapy, 3 received salvage radiotherapy, and 2 received palliative radiotherapy or chemotherapy. According to TNM stage and treatment models, further analysis was performed.For patients with stage I-II, 5-year OS, LRRFS, and DMFS for those who underwent radiotherapy (including C + RT and S + RT groups) were 83.6%, 75.0%, and 73.9%, and for those who did not receive planned radiotherapy (S + Sa group), 5-year OS, LRRFS, and DMFS were 62.5%, 12.7%, and 51.4%, respectively.For patients with stage III-IV, 5-year OS, LRRFS, and DMFS for radiotherapy group (including C + RT and S + RT groups) were 60.7%, 77.8%, and 63.2%, and for S + Sa group, 5-year OS, LRRFS, and DMFS were 57.1%, 50.0%, and 46.7%, respectively.As shown in Fig. 2, the combination of RT with chemotherapy or surgery elicited superior local control whether the tumor stage was early (p = 0.015) or late (p = 0.006), but did not significantly improve the overall survival or distant metastasis-free survival. To explore the role of systemic therapy, we performed further analysis on patients with stage III-IV.The individuals were re-grouped according to the application of systemic therapy, including adjuvant or neoadjuvant chemotherapy.5-year OS, LRRFS, and DMFS for patients who received systemic therapy (C + RT group) were 60.6%, 78.2%, and 71.2%, and for the non-systemic therapy group (S + RT or S + Sa), 5-year OS, LRRFS, and DMFS were 60.3%, 69.6%, and 51.5%, respectively.Although there was no statistical difference (p = 0.075), patients who received systemic therapy had better DMFS than those did not receive systemic therapy (Fig. 3). Failure patterns Totally, 44 of the 93 patients experienced treatment failure, and the failure patterns are shown in Fig. 4A.Local recurrences were found in 21 patients and regional failure occurred in 11 patients.The most common sites of nodal failure were level II, followed by level III.34 patients developed distant metastasis, the common sites were liver, lung, bone, and central nervous system.In addition, we analyzed the patterns of treatment failure in patients who experienced different treatment options (Fig. 4B).Patients who received planned S + RT or C + RT had obviously lower loco-regional control failure (LRF) than S + Sa group (23.5% and 15.0% vs 63.2%, p < 0.001).The difference in distant metastasis (DM) between the three groups did not reach statistical significance (p = 0.129). Prognostic analysis The results of univariate analyses showed that differentiation degree had significant effect on OS and DMFS rather than LRRFS.Treatment regimen was a significant prognostic factor affecting LRRFS (Table 2).The multivariate analyses (Table 3) indicated that there was no independent predictor of overall survival.Among the various treatment modalities, S + Sa was used as a reference, and both the C + RT group Discussion HNNENs remain a rare disease and relevant reports are primarily case reports.The present study included 93 cases of HNNENs, which constitutes one of the largest single-institution cohorts of HNNENs patients. Most epidemiological and clinical data in our study are comparable to those reported in previous studies.As described in previous studies, males have been more likely to suffer from this disease than females, the male-female ratio ranged from 1.5:1 to 8:1 (Bal et al. 2022;Ghosh et al. 2015;Issa et al. 2021;Kao et al. 2012;Mitchell et al. 2021;Pointer et al. 2017;Yu et al. 2020).The ratio in our study is 3.2:1, which is consistent with previous reports.The median age at diagnosis ranges from 57 to 64 years (Bal et al. 2022;Ghosh et al. 2015;Issa et al. 2021;Kao et al. 2012;Mitchell et al. 2021;Pointer et al. 2017;Yu et al. 2020).In our study, most of the HNNENs were located in the nasal cavity and paranasal sinus (45/95), followed by the larynx (23/95).This result is consistent with a previous study based on the Surveillance, Epidemiology, and End Results (SEER) database from 1973 to 2012 (Kuan et al. 2017).Meanwhile, there are some contrasts as well.Another study based on the SEER database analyzed a total of 789 cases of small cell carcinoma of the head and neck from 1973 to 2016, reporting that the highest incidence sites were salivary glands (over 25.0%) and nasal cavity and paranasal sinus (18.8%) (Yu et al. 2020).One study on the NCDB database analyzed a total of 1042 cases of small cell carcinoma of the head and neck from 2004 to 2012, the high-incidence areas were as follows: larynx (34.9%), nasal cavity and paranasal sinus (30.0%), and oropharynx (12.3%) (Pointer et al. 2017).We believe that differences in race may also contribute to this result.The symptoms of HNNENs are usually non-specific and mostly depend on the site of the primary tumor.Thus, the final diagnosis depends on pathological characters and genomic features (Ohmoto et al. 2021;Perez-Ordoñez 2018). Survival rate of HNNENs reported in previous literature has great differences and is closely related to the degree of differentiation.Despite that in the latest WHO 2022 classification, the terms "typical carcinoid" (welldifferentiated) and "atypical carcinoid" (moderately differentiated) were de-emphasized and used as synonyms only, and they were the preferred terminologies in previous researches (Mete and Wenig 2022;Perez-Ordoñez 2018).Patients with typical carcinoid of head and neck usually had very good prognosis, and the 5-year disease-specific survival rates of currently reported cases are almost 100% in previous reports (Kao et al. 2012;van der Laan et al. 2015).For atypical carcinoid, the 5-year survival rates reported by previous studies ranged from 46 to 83.3% (Ferlito et al. 2006;Kao et al. 2012;van der Laan et al. 2015).Moreover, HNNECs are more aggressive and had worse prognosis, with higher local recurrence rate and distant metastasis rate, reducing the 5-year overall survival rate to the range of 5-34% (Ghosh et al. 2015;Kao et al. 2012;Lin et al. 2019;van der Laan et al. 2015;Yu et al. 2020). We observed that 5-year OS in our study was higher than historical reports, which could be due to differences in treatment strategies and radiation dose or the small number of patients included in many of the previous studies. The median CTV dose in our study was 60.06 Gy, which is higher than previous studies.And as reported, the radiation dose more than 40Gy was associated with improved OS (Lin et al. 2019).In addition, no patients had distant metastases at the time of diagnosis, which would also contribute to the higher survival rate.A study based on SEER database included 789 primary cases and reported 5-year OS rate was 26.2% (Yu et al. 2020).It was noted that almost 20% of the patients included in this study had distant metastases at the time of diagnosis, which may be the main reason of the lower survival rates. Treatment modalities for HNNENs included surgery, radiotherapy, chemotherapy, or the combination of two or more treatment types.Despite no consensus on the optimal treatment currently, it is widely recognized that treatment options largely depend on degree of differentiation and clinical stage.Typical carcinoid of the head and neck can be cured with local excision or radiotherapy (Kao et al. 2012;van der Laan et al. 2015), while the best treatment for atypical carcinoid and HNNECs remained unclear (Mitchell et al. 2021;van der Laan et al. 2015).Since surgery and radiotherapy are both local treatments, some scholars have compared them and emphasized the importance of surgery.But in the study, patients underwent surgery and adjuvant radiotherapy were also included in the surgery group (Patel et al. 2015), which may cause bias.In our study, when compared to patient-treated regimen with or without RT, RT significantly improved LRRFS both in early and late stages despite no improvement in OS and DMFS.Thus, we argued that RT plays an important role in local control of HNNENs and surgery alone cannot achieve satisfactory results, no matter early or late stage.In further analysis, systemic chemotherapy improved DMFS though the result did not reach statistically significance.Based on the above results, we believed that systemic treatment is still indispensable.Similarly, a meta-analysis (van der Laan et al. 2015) suggested that atypical carcinoid needs to extend the follow-up period to 10 years to observe late recurrences to determine optimal treatment, while the small or large cell neuroendocrine carcinoma seems to benefit more from a combination of radiotherapy and chemotherapy.Consistently, another study based on SEER database investigated the 5-year disease-specific survival (DSS) of the various treatment groups for small cell HNNECs, the 5-year DSS was 26% for patients receiving surgery alone, 32% for those treated by surgery and radiotherapy, and 40% for those undergoing surgery, radiotherapy and chemotherapy (Yu et al. 2020).But there was no statistically significant difference (p = 0.21) for the different treatment modalities (Yu et al. 2020). Previous studies have reported that distant metastasis was the main failure pattern, and liver and lung were the most common sites of metastasis (Ferlito et al. 2006;Thompson et al. 2016).Our study supported these results.In addition, we also observed relatively high rate of bone and central nervous system metastases, which might suggest a role for chemotherapy.Moreover, we found that patients who received RT had obviously lower LRF than S + Sa group.Systemic chemotherapy was related with lower DM though there was no statistical significance.This result may be related to the small sample size. The prognosis of HNNENs is related to many factors, such as patient characters and tumor variables.Independent prognostic indicator included N stage, M stage, AJCC stage, and chemotherapy had been reported in previous studies (Yu et al. 2020).Some researchers reported that primary tumor sites had important impact on prognosis, for example, it was reported that sinonasal primaries appear to have better survival compared to other primary sites within head and neck (Kuan et al. 2017). Tom et al.'s meta-analysis (van der Laan et al. 2016) concluded that differentiation grade was one of the most important predictors of survival in sinonasal neuroendocrine carcinoma and was associated with choice of treatment modality.They also found that tumor staging appeared limited in value in predicting survival (van der Laan et al. 2016).Results of univariate analysis in our study showed a degree of differentiation which had an effect on OS and DMFS rather than LRRFS.The multivariate analysis revealed that S + RT or C + RT group had obviously better LRRFS while C + RT group had better DMFS.Thus, we recommend comprehensive treatment for HNNENs and individualized treatment plans for different patients. Some limitations were noted in our study.First, it was a retrospective study which was based on data from a single hospital over a long period of time.Therefore, we cannot rule out some degree of selection bias.Second, although this study included, as we known, the largest sample size in single institution, we were still unable to conduct more detailed subgroup analysis such as different primary tumor sites or various resection margin status.We expected more studies with bigger sample size and more data in future. Conclusion Radiotherapy improved local-regional control and played an important role in the management of HNNENs.The optimal treatment regimen for HNNENs remains the combination of local and systemic treatments. Fig. 2 Fig. 2 Survivals among the patients received RT vs non-RT.A OS; B LRRFS; C DMFS of stage I-II patients; D OS; E LRRFS; F DMFS of stage III-IV patients Fig. 3 Fig. 4 Fig. 3 Survivals among the patients received systemic chemotherapy vs non-chemotherapy.A OS; B LRRFS; C DMFS of stage III-IV patients Table 1 Patients' characters p values in bold incidate the existence of statistically significant difference C chemotherapy, RT radiotherapy, S surgery, Sa salvage, T tumor, N lymph node, n number, NET neuroendocrine tumor, NEC neuroendocrine carcinoma Table 2 Univariate analysis p values in bold incidate the existence of statistically significant difference OS overall survival rate, LRRFS locoregional relapse-free survival free rate, DMFS distant metastasisfree survival rate, C chemotherapy, RT radiotherapy, S surgery, Sa salvage, T tumor, N lymph node, NET neuroendocrine tumor, NEC neuroendocrine carcinoma Table 3 Results of COX multivariate analysisp values in bold incidate the existence of statistically significant difference OS overall survival rate, LRRFS locoregional relapse-free survival free rate, DMFS distant metastasis-free survival rate, M male, F female, HR hazard ratio, CI confidence interval, C chemotherapy, RT radiotherapy, S surgery, Sa salvage, Sa neuroendocrine tumor, NEC neuroendocrine carcinoma	2024-06-06T06:17:19.702Z	2024-06-04T00:00:00.000	{ "year": 2024, "sha1": "4d00f268c4fedec67da1a755a385be9653617574", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/s00432-024-05726-1.pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "f9447ced1b33db3cf21e993ebf60f83aea2f3dd5", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
86553895	pes2o/s2orc	v3-fos-license	MACRONUTRIENT ACCUMULATION AND PARTIONING IN FERTIGATED SWEET PEPPER PLANTS Information on nutrient demand during each growth stage is essential for efficient application of nutrients. A pot experiment was carried out with a Typic Hapludox under greenhouse conditions in Botucatu, SP, Brazil, aiming was to determine nutrient uptake and partition of sweet pepper plants, cultivar Elisa in randomized block design with four replications. The fertigation was simulated through 2-L PET bottles (neck down with a tube and a flow regulator at the end, simulating a drip irrigation system). Four plants per replication were collected at eight growth stages (0, 20, 40, 60, 80, 100, 120 and 140 days after the seedling transplant DAT). The period of largest extraction of nutrients for the plant occurred from 120 to 140 DAT, which coincides with the highest accumulation of dry phytomass. The highest Mg and Ca accumulation occurred in the leaves, while N, K, S and P were mostly accumulated in the fruits. Only 8 to 13% of the total amount of the accumulated macronutrientes at 140 DAT were absorbed up to the 60th DAT. Between the 61 and 100 DAT, K was the most absorbed macronutrient (60% of the macronutrients accumulated during the whole cycle). P, Ca and S were the most absorbed nutrients at the end of the cycle. Considering rates (g per plant), the most absorbed macronutrients were: N (6.6) > K (6.4) > Ca (2.6) > Mg (1.3) > S (1.1) > P (0.7). INTRODUCTION Sweet pepper (Capsicum annuum L.) is one of the most consumed vegetables in Brazil, and is better adjusted and most frequently produced under greenhouse conditions (Melo, 1997).Cultivation of sweet pepper in greenhouses demands intensive use of agri-inputs and labour.Moreover, this activity is part of a dynamic market in which seasonal prices fluctuate, demanding from the producer rigorous planning for both crop production and technology in order to decrease production risks and increase profits.Fertigation is one of the techniques utilized to achieve this goal.It consists of adding nutrients according to plan needs to the irrigation water, particularly through sprinkler and drip systems (Villas Bôas, 2001).Nutrient absorption depends on the plant phenological stage and is enhanced during flowering, fruit formation and growth (Silva, 1998).Information on these absorption rates improves the fertigation efficiency (Bar-Yosef, 1999). Sci. Agric.(Piracicaba, Braz.), v.61, n.1, p.62-68, Jan./Fev.2004 For planning fertilization it is also very important to be aware of new genotypes and growth features in different conditions, since dry matter and nutrient accumulations are closely associated to each other, so that one deficit harms other, directly or indirectly (Benincasa, 1988).The optimal nutrient absorption curve (nutrient accumulation curve) determines both the rate for application of a specific nutrient to avoid possible deficits or luxury uptake, and the adequate proportion of nutrients.One of the first studies on the nutrient absorption curves in Brazil was carried out by Haag et al. (1970) for sweetpepper (variety "casca dura") grown in nutrient solution.Further studies on nutrition, however, are necessary as a consequence of constant genetic changes of the crop, made to improve quality (Fernandes & Haag, 1972). Application of adequate nutrient rates decreases salinity problems which otherwise could negatively affect production in closed crop systems (Raij, 1993).The aim of the present study was to determine nutrient accumulation and partition along the whole cycle of a fertigated Elisa sweet pepper crop grown in greenhouse, as a support for the improvement of fertilization programs. MATERIAL AND METHODS The experiment was carried out inside a greenhouse (plastic tunnel; 20 m long, 7 m wide, and 2.5 m high) in Botucatu, SP, Brazil (22º51'S, 48º26'W; altitude 786 m).Air temperature and relative humidity were recorded by a portable thermo hygrograph placed 1.5 m above ground. Based on Raij et al. (1996), after adding dolomite lime (86% Total Neutralizing Relative Power) until the BS reached 80%, each pot received 150 g P dm -1 of soil, as thermal phosphate (17% P 2 O 5 , 20% Ca, 9% Mg and 6% S), 117 mg K as KCl, and also 150 g of an organic compound.Pots were watered to 70% of the volume corresponding to the 0.033 MPa matric potential covered with a plastic sheet and maintained for 30 days. Each pot received two seedlings of Elisa sweet pepper (Capsicum annuum L.), cultivated in trays containing commercial substratum, on July 19 th , 1999 (66 days after sowing).A distance of 0.24 m between seedlings within rows and 1.30 m row spacing were used.The development of the sweet pepper plants was limited to three stems and then the first flower (relative to the first internode) was removed.Structures for plant sustentation and growth were set up.To simulate a local irrigation system, a 2-L PET bottle was used for each pot.This bottle was suspended upside down, with removed bottom to allow inflow of water and dissolved fertilizers.A pipe was connected to the PET bottle cap, which had a water-flow controller to allow a dripping rate of 2 L h -1 .Drippers were installed 10 cm above the plant base. Nutrient rates applied through fertigation were strictly related to levels used in the field and were applied after the 54 th day after transplantation (DAT).The fertigation scheme consisted of: 1.9 g calcium nitrate diluted into 250 mL of water per plant (0.285 g nitrogen and 0.361 g calcium) up to the 66 DAT.After the fourth application (66 DAT), the level of nitrogen was increased to 0.392 g per plant (2.72 g calcium nitrate) and then the potassium (KCl) application was initiated (0.295 g K per plant).During the 10 th and 14 th applications (102 and 122 DAT, respectively), boron (0.150 mg B plant -1 ) and manganese (0.132 mg Mn plant -1 ) were added in the form of borax and manganese sulphate, respectively.N and K were applied every 4 days, (16 and 13 applications, respectively), until 133 DAT, corresponding to a total of 5.95 g N and 3.84 g K plant -1 . A randomized block design with four replications and eight sampling dates was used.Samples of plants were taken every 20 days, from day 0 to 140 DAT.In each of these eight samplings dates, four plants were collected per repetition (two pots with two plants each). Plants were collected by cutting their base close to the ground.Plant height, first-bifurcation height and number of leaves and bifurcations were recorded in the laboratory.Fruit collections started at 80 DAT, and the fruits produced by the plants that would be collected later (at 100, 120, and 140 DAT) were removed from the plant for fresh and dry matter determinations (these data were included for determination of the plant dry matter at the collection moment).Fruits larger than 10 cm in length or 7 cm in diameter were classified as commercial fruits, and smaller ones as non-commercial. Plants were washed and taken apart into individual leaves, stems, roots and fruits, and then dried in a forced-air oven at 65 o C. Each dried plant organ was weighed (dry matter), ground and submitted to chemical analyses according to Malavolta et al. (1997).Based on dry matter and nutrient concentrations, absolute and relative growth rates and nutrient levels were determined for each plant organ. Mean and standard error were calculated for each variable.The levels of plant-accumulated macronutrients from 40 to 140 DAT were analyzed by linear regression using the software SigmaPlot for Windows v.8.0 (SPSS Inc.); plant age (DAT) was the independent variable.Models (equation) were selected for best statistical adjustment (F test, 5%), best fit (R 2 ), and biological significance. RESULTS AND DISCUSSION Accumulation of dry matter by sweet pepper was slow until 60 DAT and then increased as the fructification began (Figure 1).Haag et al. (1970), Fernandes & Haag (1972) and Fontes & Monnerat (1984) observed slow growth until 75 DAT, but they used varieties rather than hybrid genotypes. The maximal growth of sweet pepper was not reached until 140 DAT (Figure 1), in contrast to reports by Crespo-Ruiz et al. (1988) who found the maximal dry matter accumulation point at 100 DAT for cv Cubanelle in Puerto Rico.After 120 DAT, the plant dry matter accumulation was greatly increased, probably because of the great fruit harvest (60% of the total commercial fruit production), which occurred up to 110 DAT (Table 1), thus reducing plant drains (Taiz & Zieger, 1991).These processes may have stimulated new vegetative growth (root, leaves and stem) that was indicated also by increased plant height and number of bifurcations and leaves (Table 2).Villas Bôas ( 2001) studied this same hybrid under similar conditions and found taller plants (109 cm) with fewer leaves (55) and more bifurcations (34) at 174 DAT.Panelo (1995) reported that the Vidi hybrid had the lowest plant length (50 to 70 cm) after two month of ongoing harvest (~ 143 DAT), while Elisa hybrid presented intermediate length (80 to 90 cm) at this time, in comparison to the other studied hybrids. The first-bifurcation height reached 19 cm at the end of the observations.This may depend on the cultivation system (Faria Jr., 1997), but may also be altered by the number of conducted stalks (Tivelli et al., 1998).An excess of fertilizers, mainly nitrogenous, may promote excessive plant growth and, consequently, the first bifur-cation.For Elisa hybrids of the same age, Tivelli (1999) found 22.4 cm height for the first bifurcation. After transplantation, the plant remains in an adjustment period up to 20 DAT, named "establishment phase" (Crespo-Ruiz et al., 1988), during which dry matter is accumulated mostly in the leaves.After that (from 20 to 60 DAT), dry matter is allocated to the stem and roots (Figure 2), enlarging the radicular system.Dry matter accumulated in the leaves decreased from 64.5% at 1 harvesting either without ripe fruits and without commercial fruits.20 DAT to 20.4% at 100 DAT (Figure 2), probably as a function of dry matter allocation to the stem and root until 60 DAT and then to fruit formation (from 60 to 100 DAT). The most intense fructification from 80 to 100 DAT increased dry matter in fruits (54.2% at 100 DAT) instead of leaves, stems and roots.During this period, the number of bifurcations did not increase, but plant height and number of leaves increased slightly (Figure 2).At the fructification period (60 to 100 DAT), the absolute growth rate (AGR) of plants increased from 0.6 to 1.8 g plant day -1 , mainly because fruits increased from 0.7 to 14 g plant day -1 (Figure 3). Plant AGR decreased slightly (1.6 g plant day -1 ) from 100 to 120 DAT, followed by a high increase (6.2 g plant day -1 ) from 120 to 140 DAT, probably as a result of generalized growth of all plant organs.The maximal AGR found by Dias (2000) for the hybrid Elisa until 224 DAT (6.85 g plant -1 ) was similar to that herein described.The dry matter distribution at 140 DAT was similar to that observed at 80 DAT when harvesting was initiated, with about 70% of dry matter accumulated in roots, leaves and steams (Figure 2). The period of highest nutrient extractions occurred from 120 to 140 DAT, coinciding with the greatest accumulation of dry matter (Figure 1).During this period the accumulated N levels doubled from 3 to 6 g N day -1 (Figure 4).On the other hand, absorption of P, Mg and S increased only after 80 DAT, while absorption of N, K and Ca increased after 60 DAT (Figures 4 and 5).The higher accumulation of Mg and Ca occurred in the leaves, but N, K, S and P were mostly accumulated in the fruits (Tables 3 and 4). The accumulation curves plotted in Figures 4 and 5 allow the estimation of the amount of extracted nutrients by the plants during each developmental phase, starting from 40 DAT.This indicate the need of fertilizer per fertigation, which for sweet pepper crops should begin between at 40 to 50 DAT, since fertilization during the transplantation period, made mainly with manure, meets plant demands efficiently (Villas Bôas, 2001). N and K were the most extracted macronutrients, with similar rates of extraction for N, followed by Ca, Mg, S and P in decreasing order of concentration (Table 3).The levels of extracted nutrients found by Silva (1998) for the Mayata hybrid sweet pepper (N = 5.6 g; P = 0.5 g; K = 10.0 g; Ca = 3.5 g; Mg = 1.7 g and S = 1.0 g) were similar to those observed here, except for K, which was 56% higher.Haag et al. (1970), Santiago & Goyal (1985), Graifenberg et al. (1985), Dias (2000) and Villas Bôas (2001) also reported greater extraction of K in comparison to N. This might be a consequence of using different cultivars and/or luxury uptake, since in the present study there was no K deficiency; and K levels in the leaves are within the range considered normal by Trani et al. (1996), or even above this range, as considered by Malavolta et al. (1997) (Table 5).According to these two later studies, the levels of the other macronutrients are within the adequate range for sweet pepper at 100 DAT (complete flowering), which is the indicated period for sampling leaves for chemical analyses. Table 3 -Accumulation of nutrients (mean ± SE) in total dry matter and plant parts of fertigated sweet pepper, according to days after transplantation (DAT). (1) Because of insufficient material, samples of blocks were mixed for analyses, not allowing mean and SE evaluations.SE = standard error P was the most extracted element by the fruits, followed by N, S and K at similar percentages (Table 6).On the other hand, Ca and Mg were the nutrients least extracted by the fruits.Dias (2000) found the same macronutrient exportation order for the Elisa sweet pepper: P=60%; N=40%; K=40%; S=35%; Ca=24% and Mg=18%. From the total amount of macronutrients accumulated until 140 DAT, only 8 to 13% were absorbed up to 60 DAT (Table 7).Between 61 and 100 DAT, K was the most absorbed macronutrient (60% of the total accumulated in the cycle); P, Ca and S were more absorbed at the end of the cycle (from 101 to 140 DAT).Production of commercial fruit of green sweet pepper up to 140 DAT reached 1.3 kg per plant (Table 1).Tivelli (1999) found similar results for Elisa hybrids cultivated in a closed system, which reached about 1.1 kg per plant.Higher yield in commercial fruits was ob-SE = standard error Table 4 -Accumulation of nutrients (mean ± SE) in dry matter of total plant and plant parts of fertigated sweet pepper, according to days after transplantation (DAT). tained as a result of larger space between plants (0.48 x 1.30 m) (Villas Bôas, 2001) and because of the extended fruit-harvesting period (5 months) (Melo, 1997).In these cases, the commercial fruit yield reached 2.5 and 3.85 kg per plant, respectively. Figure 1 - Figure 1 -Accumulation of dry matter (total and for each organ) by sweet pepper fertigated at different days after transplantation (DAT). Figure 2 - Figure 2 -Percentual distribution of dry matter in each organ of sweet pepper fertigated at different days after transplantation. Figure 3 -Figure 4 - Figure3-Absolute growth rate (AGR) for the whole plant(total) and for each plant organ for sweet peppers fertigated at different days after transplantation. Figure 5 - Figure 5 -Adjusted curves for accumulation of P, Mg, and S in sweet pepper plants fertigated from 40 to 140 days after transplantation.*significant R 2 by F test at α = 5%. Table 1 - Sweet pepper fruit production performance according to days after transplantation. Table 2 - Means (± SE) of plant and first-bifurcation heights, number of bifurcations and leaves according to days after transplantation. Table 5 - Concentration of nutrients in leaves (mean ± SE), according to the cycle of fertigated sweet pepper. Table 6 - Quantities of nutrients extracted by the plant and exported by fertigated sweet pepper during the 140-day cycle.	2019-01-03T22:43:56.469Z	2004-02-01T00:00:00.000	{ "year": 2004, "sha1": "4776135098aac31cd035efef951a91653e29cb74", "oa_license": "CCBY", "oa_url": "https://www.scielo.br/j/sa/a/9hbSfWrs8rZj4VTY68nLSzs/?format=pdf&lang=en", "oa_status": "GOLD", "pdf_src": "ScienceParseMerged", "pdf_hash": "4776135098aac31cd035efef951a91653e29cb74", "s2fieldsofstudy": [ "Agricultural And Food Sciences" ], "extfieldsofstudy": [ "Biology" ] }
56459110	pes2o/s2orc	v3-fos-license	Non-Universal Soft SUSY Breaking and Dark Matter An analysis is given of the effects of non-universal soft SUSY breaking masses in the Higgs sector and in the third generation squark sector, and it is shown that they are highly coupled. Analytic expressions are obtained for their effects on the parameters $\mu,m_A$ and on the third generation squark masses. Non-universality effects on dark matter event rates in neutralino-nucleus scattering are analysed. It is found that the effects are maximal in the range $m_{\tilde\chi_1}\leq 65$~GeV where the relic density is governed by the Z and Higgs poles. In this range the minimum event rates can be increased or decreased by factors of O(10) depending on the sign of non-universality. Above this range Landau pole effects arising from the heavy top mass tend to suppress the non-universality effects. The effect of more precise measurements of cosmological parameters on event rates, which is expected to occur in the next round of COBE like sattelite experiments, is also investigated. Implications for the analysis for dark matter searches are discussed. Introduction Although there is currently a great deal of evidence for the existence of dark matter in the universe [1], the nature of such dark matter is still uncertain. COBE and other experiments have provided further insights into the nature of dark matter which strongly suggests more than one component to its structure [2]. Thus the simplest types of models to fit the power spectrum seen by COBE and other experiments require two components, a hot component and a cold component. Several other possibilities have also been discussed such as those using a cosmological constant. In our analysis here we shall adhere to the simple two component picture of dark matter, with a hot component(which could be a light neutrino with mass in the range of a few electron volts), and a cold component which we assume is the lightest supersymmetric neutral particle, the neutralino. (For a review of supersymmetric dark matter see ref. 3). For the purpose of our analysis we assume a mix of hot and cold dark matter in the ratio of Ω HDM : Ω CDM = 1 : 2 (consistent with the COBE data), where Ω i = ρ i /ρ c where ρ i is the matter density in the universe due to matter of type i and ρ c is the critical matter density needed to close the universe. Assuming an inflationary scenario with Ω = 1, the condition that Ω B ≤ 0.1 as implied by the Big Bang Nucleosynthesis, and h in the experimental range 0.4 ≤ h ≤ 0.8 where h is the Hubble parameter in units of 100 km/s Mpc, one finds Ωχ 1 h 2 in the range 0.1 ≤ Ωχ 1 h 2 ≤ 0.4 (1) Ωχ 1 h 2 is the quantity computed theoretically and thus the above condition acts as the primary constraint on the neutralino dark matter analysis. We shall discuss in the latter part of this paper the effect of constraining the Ωχ 1 h 2 range more narrowly. The purpose of this paper is to analyse the effects of non-universal boundary conditions on soft SUSY breaking masses at the GUT scale on event rates in neutralino-nucleus scattering, as almost all of the previous analyses, with the exception of the work of Ref. (4), are in the framework of supergravity unification using only the universal boundary conditions. The outline of the paper is as follows: in Sec. 2 we give the salient points of supergravity unification which provides the framework of our analysis. In Sec. 3 we discuss non-universal soft breaking in the Higgs sector and in the third generation squark sector and show that their effects at low energy are highly coupled. We then obtain solutions in a closed form at the electro-weak scale of the relevant low energy parameters that contain the non-universalities. In Sec. 4 we discuss the constraints that are imposed on the event rate analysis. In Sec. 5 we give the basic formulae that enter in the analysis of scattering of the neutralino by nuclear targets. In Sec. 6 results of the event rate analysis are given and a comparison made between the cases with universal and nonuniversal soft SUSY breaking boundary conditions at the GUT scale. In Sec. 7 we discuss the effects of grand unification on non-universal parameters. In Sec. 8 we discuss the effects that a more accurate determination of the Hubble parameter, which is expected to occur in the next round of COBE like sattelite experiments, will have on the event rate analysis. Conclusions are given in Sec. 9. New results of this paper are contained in Secs. 3,6,7,8, in Tables 1-5 and in Appendix A. Neutralino Dark Matter in Supergravity Grand Unification In the analysis of this paper we use the framework of supergravity grand unification(SUGRA) [5]. In this picture one assumes the existence of a supergravity grand unified theory which in its symmetric phase operates in the region between the string scale and the grand unification scale. Supersymmetry is broken in the theory via a hidden sector. In the minimal supergravity unification one finds that after breaking of supersymmetry and of the grand unified symmetry (where one assumes that the grand unified theory breaks to the gauge group SU(3)×SU(2)×U(1) ) one has a low energy theory (obtained by integrating out the heavy modes and the modes of the hidden sector) which can be given in terms of just five arbitrary parameters [5] − [8]. These consist of the universal soft SUSY scalar mass at the GUT scale m 0 , the universal gaugino mass at the GUT scale m 1/2 , the universal trilinear scalar coupling at the GUT scale A 0 , B 0 which is the co-efficient of the bilinear term µ 0 H 1 H 2 at the GUT scale, and the Higgs mixing parameter µ 0 at the GUT scale. In addition there are the GUT parameters M G and α G which are determined by the grand unification condition. A remarkable aspect of supergravity grand unification is the breaking of the electro-weak symmetry by radiative effects. Radiative breaking is governed by the equations [9] where B is the value of the parameter B 0 at the electro-weak scale, µ 2 i = m 2 H i + Σ i where m 2 H i is the running H i mass at the scale Q ≈ M Z and Σ i are loop corrections [10,11]. m 2 H i are given by Here t=ln(M 2 G /Q 2 ), and e,f,g,h,k are form factors defined in Ref. (12). On using the radiative breaking equations one can determine the parameter µ from the Z boson mass and one can also eliminate the parameter B in favor of tanβ ≡< H 2 > / < H 1 >. Thus in the low energy domain, the theory can be described by the parameters m 0 , mg, A t , tanβ, sign(µ) (5) where A t is the value of A 0 at the electro-weak scale. Further, at the electroweak scale the SU(3) × SU(2) × SU(1) gaugino masses are given by m i = (α i /α G )m 1/2 . ( For radiative corrections to this formula see Refs. (13)(14)). We note that the situation in minimal supergravity is in sharp contrast to that for MSSM. In MSSM one has 110 parameters in the low energy domain. In contrast, in minimal supergravity unification (MSGM) one has only four parameters (and one sign) as discussed above. The minimal model has 32 supersymmetric particles, and all their masses and coupling constants are determined by just the parameters of Eq. (5). One of the interesting features that emerges from supergravity unification is that the model predicts as a consquence of its dynamics which particle is the lowest lying supersymmetric particle(LSP). One finds that over almost all of the parameter space that the neutralino is the LSP. Thus with the assumption of R parity invariance, one finds that supergravity unified theory provides a candidate for cold dark matter. The LSP neutralino is an admixture of four neutral states, i.e., of gauginosW 3 ,B and of HiggsinosH 1 ,H 2 . Denoting the lightest neutralino byχ 1 we may writẽ where the co-efficients n i are to be determined by diagonalizing the neutralino mass matrix given in Appendix A. An interesting property of radiative breaking is that it produces the phenomenon of scaling in a significant part of the parameter space [15]. In the region of scaling one has µ 2 >> M 2 Z . In this region the eigenvalues and the eigenvectors of the neutralino mass matrix can be computed in a perturbative fashion by expanding in powers of M Z /µ [16]. The analysis shows that in the scaling limit the neutralino is mostly a Bino with n 2 > 0.95. Further one finds that n 3 , n 4 and n 1 are typically of first order, i.e., O(M Z /µ). These expansions are useful in understanding the relic density and the event rates. We add here a note of caution. Inspite of the fact that the Higgsino components are generally small their effects on event rates are generally very significant. Thus large inaccuracies can result from a neglect of the Higgsino components. Effects of Non-universal Soft SUSY Breaking In this section we want to study the effects of non-universalities in soft SUSY breaking on the low energy parameters. While most of the previous analyses in supergravity grand unification have been carried out with universal soft SUSY breaking, the general framework of the theory allows for non-universal soft SUSY breaking terms. Thus, for example, under the assumption of a general Kahler potential, deviations from universalities can be generated [17]. Of course the non-universalities are severely constrained, most significantly from flavor changing neutral current(FCNC) constraints. In the analysis here we shall limit the nature of non-universalities to the Higgs sector and to the third generation sector. Higgs sector non-universalities are not strongly constrained by FCNC and have been discussed in the literature [18,19,20,4] often in the context of certain SO(10) models to achieve radiative breaking of the electroweak symmetry. In our analysis here we show that the non-universalities in the Higgs sector and the non-universalities in the third generation up squark sector are highly coupled due to renormalization group effects below the GUT scale. It is convenient to parametrise the non-universalities in the Higgs sector by δ 1 , δ 2 and the non-universalities in the third generation squark sector by δ 3 , δ 4 so that at the GUT scale M G one has where m 0 is the universal scalar mass of the first two generation masses. We shall assume that the remaining sectors of the theory are universal. We shall give an analytic solution to the non-universal effects on the mass spectra at low energy. We begin by a discussion of the case of the H 1 mass and the masses of the first two generations of sparticles. For m 2 H 1 we obtain Here the last term arises from the T r(Y m 2 ) term in the renormalization group evolution equations [13]. This term vanishes in the universal case because of the anomaly cancellation constraint T r(Y ) = 0. However, for the non-universal case it is non-vanishing and one has (10) where all the masses are at the GUT scale and n g is the number of generations. In Eq. (11) p is defined by whereα 1 = g 2 1 (0)/(4π) and g 1 (0) is the U(1) gauge coupling constant at the GUT scale. The corrections to the sparticle masses for the first two generations are listed in Appendix A. We note that although the non-universalities at the GUT scale in the first two generations vanish there are effects at the electroweak scale in the sparticle masses of the first two generations because of nonuniversalities in the Higgs sector and in the third generation via the trace anomaly term. To get an idea of the size of the effects from this term, for M G = 10 16.2 GeV and α G = 1/24 one has p=0.0446. Thus the effect of the trace term is order a few percent for \|δ i \| < 1. The analysis of non-universalities on the squark masses in the third generation is more complicated. Here m 2 H 2 , m 2 U and m 2Q obey the coupled renormalization group equations (61)-(63) of Appendix A. Solution to these gives for m 2 where ∆ H 2 is given by and D 0 defines the top Landau pole, i.e., Here y 0 = h 2 t /(4π) 2 where h t is the top Yukawa coupling, and In Eq. and m 2 t L , m 2 t R are given by and Here m 2Q and m 2 U obey the set of coupled equations given in Appendix A. With the assumption of non-universalities of Eqs. (7) and (8) Similarly for m 2 U one has In the above, the non-universalities are all contained in the quantities ∆ H 2 , ∆Q, ∆Ũ and in the corrections involving the term S 0 p. We note that because of a peculiar accident that the sum of the corrections proportional to S 0 p in the subsectors involving H 2 ,Q andŨ vanishes, the trace anomaly receives no top quark Landau pole enhancement. A proof of this result is given in Appendix A. An analysis similar to the above holds in the bottom squark sector. Details are given in Appendix A. In addition to above, the quantities µ and m A are also effected. Using the radiative breaking relation Eq. (2) we find the following closed form solution for µ 2 to one loop order: Here where t ≡ tanβ and where the functions e,f,g,k are as defined in Eqs. (3) and (4). Similarly, for m 2 A one has where In the above we have assumed the existence of non-universalities on phenomenological grounds. Theoretically there are several possible sources from where such non-universalities can arise. One source of non-universalities is the general Kahler potential. In general the Kahler potential can have generational dependent couplings which lead to non-universalities in the visible sector after the breaking of supersymmetry in the hidden sector. However, even if the couplings in the Kahler potential were generation blind and the soft SUSY breaking masses for the scalars were universal at a scale higher than the GUT scale one will have non-universalities at the GUT scale arising from renormalization group running of the soft SUSY breaking parameters. Thus, for example, if one has universality of the soft parameters at the string scale, the running of these parameters from the string scale down to the GUT scale will generate non-universalities at the GUT scale. Model calculations indicate that such running typically generates \|δ 1,2 \| ≤ 0.5, and a similar level of non-universality in the relevant mass spectra at the electro-weak scale is expected. However, enhancement of non-universalities can occur under special circumstances. To show this enhancement we consider the expression for µ 2 . Here the non-universalities are all contained in C 1 . If the dominant universal terms cancel, then the non-universal effects become large. This situation can be easily seen to arise for large tanβ. Here for the case when m t ≈ 167 GeV, the universal part cancels since D 0 ≈ 1/3 and thus the non-universality effects get enhanced. A similar enhancement of non-universalities also occurs from the A 0 dependence, although in a somewhat different manner. Here the enhancement occurs when the residue of the Landau pole in A 0 vanishes. All three quan- where (12), one finds that The Landau pole contribution vanishes when A R = 0 which occurs when A t ≈ 0.61mg [21]. We give a graphical description of this phenomenon in Fig. 1 where µ is plotted as a function of A t . For the specific set of parameters chosen in Fig. 1 the residue of the Landau pole vanishes when A t /m 0 ≈ 0.7. One finds that Landau pole effects are suppressed close to this value of A t , and there is considerable dispersion among the three curves corresponding to the sets of values δ 1 = 0 = δ 2 , δ 1 = 1 = −δ 2 and δ 1 = −1 = δ 2 . However, away from the region of A t /m 0 ≈ 0.7 one finds that the Landau pole effects begin to dominate and diminish the effect of non-universalities. Constraints on Dark Matter We explore the parameter space of supergravity unified models under the naturalness constraints which we assume to imply The constraints of electro-weak breaking and the condition that there be no tachyons then imply the following range for A t [21,16]: Constraints also arise from imposition of the condition that there be no nearby minima with lower energies that break color and charge conservation [22](CCB constraints), and from experimental lower limits on the SUSY mass spectra given by CDF, D0 and LEP experiments. In addition to the above there is an important constraint that arises from the decay b → s + γ. This decay proceeds via loop corrections and is thus sensitive to physics beyond the Standard Model(SM). For the case of the SM, the loop contribution involves a W-t exchange, while for the supersymmetric case one has additional contributions arising from W − H − ,W −t, andZ −b exchanges. Thus SUSY contributions are a priori as important as the SM contributions, and so one expects b → s + γ decay to act as an important constraint on the parameter space of supergravity models. The CLEO value for the branching ratio for B → X s + γ is [23] If one combines the errors in quadrature then BR(B→ X s γ) ∼ = (2.32 ± 0.67) × 10 −4 . The SM prediction gives BR[B → X s γ] ∼ = (3.28 ± 0.33) × 10 −4 for m t = 174 GeV [24,25,26]. In supersymmety, theoretical analyses when subject to the current experimental limits, impose serious constraints on the parameter space of the theory. The parameter space after the above constraints are imposed must still be subject to the constraint of relic density of Eq. (1). The quantity Ωχ0 1 h 2 is computed theoretically in each point in the supergravity parameter space using the relation [28,29] Ωχ0 where T γ is the current background temperature, T f is the freezeout temperature, (Tχ0 1 /T γ ) 3 is the reheating factor, N f is number of massless degrees of freedom at freezeout, Here < σv > stands for the thermal everage, where σ is the the neutralino annihilation cross-section and v is the neutralino relative velocity. In the analysis of the relic density we have used the accurate method for its computation [30,31,32,33] which carries out a correct thermal averaging over the Z and the Higgs poles which appear in the s channel in the neutralino annihilation. The accurate method for the computation of relic density is important as it has significant effects in the event rate analysis. The relic density computed in the above fashion is then subject to the constraints of Eq. (1) which further limits the parameter space of the theory. Detection of Neutralino Dark Matter We discuss now the possibilities for the detection of the neutralino dark matter. Several techniques, both direct and indirect, have been discussed over the years for its detection.The direct method includes (i) scattering by nuclei [34]- [43] in terestial detectors, (ii) scattering by bound electrons [44], (iii) use of overheated microbubbles [45], while the indirect methods include, (iv) annihilation in the center of sun and earth [46,47], and (v) annihilation in the halo of galaxies. For our analysis here we limit ourselves to consideration of case (i). The total interaction in the scattering of neutralinos by quarks is given by where P R,L = (1 ± γ 5 )/2. The part with terms A L , A R is the spin dependent(SD) part while the remainder is the spin independent(SI) part. The total event rate can thus be written as follows: where ρχ 1 is the local mass density ofχ 1 incident on the detector, and vχ 1 is the incidentχ 1 velocity. R SI , is given by and the spin dependent rate is given by where J is the nuclear spin and λ is defined by < N \| → Si \| N >= λ < N \| → J \| N >. A SI and A SD are the corresponding amplitudes. We note that for large M N , R SI ∼ M N while R SD ∼ 1/M N . These results imply that the spin independent scattering becomes more dominant as the nucleus becomes heavier. Both R SI and R SD contain theoretical and experimental uncertainties. A major source of uncertainty in R SI arises from the uncertainty in the matrix elements of the operator m qq q between nucleon states. There can be as much as 50% uncertainty in the strange quark contributions [48] leading to a 30% uncertainty in R SI . For the case of R SD the major source of ambiguity arises from the matrix elements of the axial current between nucleon states, i.e., of ∆q defined by < n\|qγ µ γ 5 q\|n >=2s µ n ∆q, where s µ n is the nucleon spin 4-vector. There exist two sets of determinations of ∆q, an old determination using the EMC data [49] and a new determination using the SMC data [50]. For most nuclei the difference is not major and in our analysis here we use the values of ∆q from the newer determination [50]. In addition both rates have uncertainties due to the nature of the nuclear form factors. Thus theoretical predictions are accurate to within perhaps a factor of two. Event Rates with Universal and Non-universal Soft Breaking We give now an analysis of the event rates and draw a comparison between the case where one uses universal boundary conditions at the GUT scale for the soft SUSY breaking parameters and the case when on has non-universal boundary conditions at the GUT scale. We begin with a discussion of event rates when one includes non-universalities in the Higgs sector, i.e., when δ 3 = 0 = δ 4 . From Eqs. (23) and (24) one can see that a positive δ 1 and a negative δ 2 make a positive contribution to µ 2 while a negative δ 1 and a positive δ 2 make a negative contribution to µ 2 . If one limits δ 1 and δ 2 so that \|δ i \| ≤ 1(i=1,2), then the case δ 1 = 1 = −δ 2 gives the largest positive contribution to µ 2 while the case δ 1 = −1 = −δ 2 gives the largest negative contribution to µ 2 . These cases represent the extreme limits of how large the non-universality efffects can be within our prescribed limits on δ i . With the above in mind we focus on three cases to get an idea of the differences in the event rates between the universal and the non-universal cases: Fig. 2 we exhibit the maximum and minimum of event rates for xenon for the three cases listed above for µ > 0.The analysis is done by mapping the full parameter space limited only by the naturalness constraints of Eq. (29), the constraints on the sparticle masses from CDF, DO and LEP experiments, the relic density constraint of Eq. (1) and the experimental constraint from b → s + γ of Eq. (31). The dips in the region below mχ 1 ≤ 65 GeV arise due to the rapid annihilation in the vicinity of the Z pole and the Higgs pole. We note that the accurate method mentioned earlier for the computation of the relic density is important for the correct analysis of the event rates in this region. Comparison of cases(i) and (ii) shows that the event rates of (i) below the region mχ 1 ≤ 65 GeV can be enhanced by a factor of O(10) or more in the minimum curves for case(ii).This effect can be understood as follows: for case(ii) one has δ 1 < 0 and δ 2 > 0, which according to Eq. (23) drives µ 2 towards the limit µ 2 < 0 eliminating such points from the parameter space. Now this effect is more important for small tanβ. Since small tanβ governs the minimum event rates one finds that in this case since small tanβ tend to get eliminated one has the effect that the lower limit of the event rates are raised. In contrast, for case(iii) one has δ 1 > 0 and δ 2 < 0 which gives the opposite effect as expected, i.e., on finds that the minimum event rates are reduced by a factor of O(10). However, above mχ 1 ≥ 65 the non-universalities do not dramatically affect the event rates and the event rates have a uniform fall off as a function of the neutralino mass for all the three cases. A similar effect occurs for larger values of the top quark mass. However, here one finds that a larger value of the top mass gives a larger Landau pole contribution which diminishes the relative contribution of the non-universal terms. We note that the spin independent interaction contributes significantly more than the spin dependent intereaction. This result holds not only for a heavy target such as xenon but also for retatively light targets such as CaF 2 . In Tables 1, 2 and 3 we give a comparison of event rates for six additional target materials , i.e., He, CaF 2 , GaAs, Ge, NaI, and Pb, for the universal and non-universal soft SUSY breaking in the Higgs sector. (Recall, however, that there are significant uncertainties discussed above in the predictions.) Phenonmena similar to the ones discussed for xenon also appear for these cases. Many of these materials listed in Tables 1, 2 and 3 are already being used as targets in the dark matter detectors. We note that for all the targets considered the event rates span several decades in magnitude within the allowed domain of the parameter space. Thus a full exploration of the parameter space certainly poses a formidable challenge to experimentalists. The analysis we have given so far is for the case µ > 0. As discussed earlier for the case µ < 0 a large part of the parameter space is eliminated because of the b → s + γ constraint. Consequently the maximum event rates for the µ < 0 cases are much smaller than for the µ > 0 case. A comparison of the allowed range of event rates for xenon for the cases µ > 0 and µ < 0 is given in Table 4 (µ > 0) and Table 5 (µ < 0). One finds that the event rates for the µ < 0 case are typically O(10 −2 − 10 −3 ) smaller than for the µ > 0 case. Another important phenonmenon is that for both signs of µ the parameter space corresponding to A t /m 0 < −0.5 is eliminated. The reason that positive values of A t are eliminated arises from the fact that the lightest stop mass turns tachyonic due to Landau pole effects, since the residue of the Landau In the analysis of this section thus far we assumed universality in the third generation sector and examined only the effects of non-universality in the Higgs sector. However, as discussed in Sec. 3 the non-universalities in the Higgs sector and in the third generation sector are highly coupled. Therefore, we turn now to a discussion of what we might expect, from the non-universality in the third generation sector. To get an idea of the nature of corrections one might expect, we display below the numerical sizes of the non-universal corrections ∆ H 2 , ∆t L , ∆t R . For values M G = 10 16.2 GeV, α G = 1/24 and m t = 175 GeV one has D 0 ≃ 0.27 and from Eqs. (13), (20) and (22) one finds From Eq. (40) we find that a positive δ 2 in ∆ H 2 can be simulated by a negative value of δ 3 or a negative value of δ 4 , and a converse situation occurs for a negative value of δ 2 . The effects of δ 3 and δ 4 on ∆t L , ∆t R are more complicated since δ 3 , δ 4 enter ∆t L , ∆t R with opposite signs. The stop masses do not enter directly in any significant manner in the amplitudes in the neutralino mass range we investigate, although they do affect the event rates by affecting the parameter space. In spite of this complexity we can glean the pattern in which δ 3 , δ 4 affect the event rates by examining the way they enter ∆ H 2 . In Fig. 3 we exhibit the event rates for xenon when δ 1 = δ 2 = δ 4 = 0 and δ 3 takes on a range of values. The solid curve in Fig. 3 is for the case δ 3 = 0, the dotted curve for the case δ 3 = 1, and the dashed for the case δ 3 = −1. Fig. 3 shows that the minimum event rates in the region below mχ 1 < 65 GeV are lower relative to universal case for the case δ 3 > 0. This is as expected from our general analysis above, i.e., it is similar to the case δ 2 < 0 in Fig. 2. Similarly the minimum event rates in the region mχ < 65 GeV are enhanced relative to the universal case for δ 3 < 0 which parallels the case δ 2 > 0 for Fig. 2. This is again as expected. An analysis similar to the above for the case when δ 1 = δ 2 = δ 3 = 0 and δ 4 takes on a range of values is given in Fig. 4. The result of non-universalities is qualitatively similar to that of Fig. 3 as expected. We note, however, that the parameter space allowed by the non-universalities in the third generation would be somewhat different and so one does not have a complete correspondence between the non-universalities in the Higgs sector (i.e., δ 2 ) and those in the third generation sector (i.e., δ 3 and δ 4 ). Effects of Grand Unification On Non-Universal Parameters In the previous section, the parameters describing non-universal soft SUSY breaking masses, i.e., δ 1 , .., δ 4 , were viewed as independent quantities. This would be the case if the scale of SUSY breaking occured below M G , since there only the SM gauge group is assumed to hold. However, if the SUSY breaking scale occurs above M G (e.g., at the string scale M str ), then the soft breaking masses at M G must obey constraints imposed by the grand unification group G. We consider here this situation for the case where G contains SU(5) as a subgroup. This includes G=SU(5), SO(N) for N≥ 10 or E(6) which essentially include all the grand unification groups that have been examined in significant detail. For this class of groups, we may characterize the light matter at M G in terms of the SU(5) qunatum numbers, i.e., matter occurs in 10 a and5 a representations (a=1,2,3 is a generation index) and the light Higgs doublet in a 5 and5 representation. One has then for the sfermions [51] 10 a = (q a ≡ (ũ La ,d La ); u a ≡ũ Ra ; e a ≡ẽ Ra ) (43) 5 a = (l a ≡ (ν La ,ẽ La ); d a ≡d Ra ) (44) The soft breaking masses have the form m 2 10a = m 2 0 (1 + δ 10 a ); m 2 5a = m 2 0 (1 + δ5 a ) (45) with the Higgs masses obeying Eq. (7). There are therefore 8 independent soft breaking parameters for this general case. ( Eqs. (7, 43) contains 9 parameters δ 10 a , δ5 a , δ 1,2 , m 0 , but actually one is redundant.). Thus we have the relations For higher gauge groups, there can be additional relations. (For example, for SO(10) with non-universalities in the third generation, some models of symmetry breaking give [53] δ5=δ 2 -δ 1 .) One may impose further group constraints on the above parameters. Thus the suppression of FCNC suggests a near degeneracy in the first two generaitons, which occurs naturally in a SU(2) H model where these generations are put into a doublet (d) representation of SU(2) H and the third generation into a singlet (s) representation [54]. Then Eqs. (45) reduce to where quantities without generation indices are SU(2) H singlets. This model would have 6 independent parameters. While the FCNC constraints do not imply that the first two generation soft mass of the 10 representation is degenerate with that of the5 representation, a further simplification occurs if one assumes that this is the case, i.e., one lets One can then absorb the (1 + δ (d) ) factor into the m 2 0 as the common mass for a=1,2, and use that as the reference mass for the third generation and Higgs masses, reducing the number of parameters to 5: δ 10 (s) , δ5 (s) , δ 1,2 and m 0 . In the notation of Sec. 7, this is equivalent to non-universalities specified by where δ 5 ≡ δ5 (s) = δb R = δ l 3 . The parameter δ 5 does not enter directly into the event rate R calculation or into the Landau pole or b → s + γ analyses. It does, however, affect R indirectly in that it will enter into the relic density calculation and affect the determination of the parameter space that obeys Eq. (1). However, there it enters only into the t-channel poles which generally make small contributions to the χ 0 1 annihilation cross section. We discuss now the effect of the model of Eq. (50) on event rates. For illustration we consider two cases with δ i (i=3,4) chosen at the extreme values under the reasonable constraint δ i ≤ 1. These are case(a): δ 3 =δ 4 =1, and case(b): δ 3 =δ 4 =-1(and δ 5 set to zero). For case(a) we find from Eqs. (13), (20) and (22) For δ 2 in the range (1,-1) one finds that ∆ H 2 lies in the range (-0.1,-1.37) while ∆Q lies in the range (0.64, 0.88) and ∆Ũ lies in the range (0.27,0.76). Because of the non-universalities arising from δ 3 = δ 4 > 0 this case is again different from the cases we have discussed before. However, we can still draw some rough comparisons with the previous cases. Thus, for example, we note that since the effective value of δ 2 in this case is negative the minimum of the event rates should be similar to the minimum dotted curve of Fig. 2 where δ 2 was also negative. The result of the analysis is presented in Fig. 5. We see that indeed the minimum of the event rates here is similar to the minimum of the event rates in Fig. 2 as expected. In general, the effect of non-universalities is not very large here. A similar analysis holds for case(b). Here again using Eqs. (13), (20), (22) and D 0 ≃ 0.27 we find that ∆Q ≃ −0.76 − 0.12δ 2 For δ 2 in the range (1,-1) one finds ∆ H 2 in the range (0.1,1.37), ∆Q in the range (-0.64, -0.88), and ∆Ũ in the range (-0.27,-0.76). We note that these ranges are exact mirrors of the ranges for case(a) but with a negative sign. Of course, the full dynamics of this system is very complicated since the allowed parameter space is affected differently from previous cases because of the specific nature of non-universalities here. However, as in case(a) we can glean some understanding by a comparison with the previous analyses. We note that the effective value of δ 2 is positive here. Thus one expects that the minimum event rates might show the same pattern as the positive δ 2 case of Fig. 2. The result of the analysis is exhibited in Fig. 6. We find that indeed the minima curves of Fig. 6 have a behavior similar to the dashed curve of Fig. 2 as expected. The remarkable feature here is the increase by a factor of 10 in the maximum event rates when δ 2 is positive. This is due to the fact that δ 2 ,δ 3 , δ 4 now act together to reduce µ 2 (see Eqs. (23) and (24)) which increases the maximum event rates 38 . 8. Effects of More Accurate Ωh 2 Determination As mentioned in the Introduction the next round of satellite experiments are expected to determine the Hubble parameter to within O(10%) or better accuracy. Such an accurate determination of the Hubble parameter will put more stringent bounds on the neutralino relic density Ωχ 1 h 2 and hence more stringently constrain the SUGRA parameter space. We investigate here the effects of these more stringent constraints on the neutralino dark matter event rates. For specificity we consider a narrow band of Ωχ 1 h 2 in the range 0.225 < Ωχ 1 h 2 < 0.275 (57) Fig. 7 gives the maximum and the minimum event rates as a function of thẽ χ 1 mass. A comparison of Figs. 2 and 7 shows that a narrower range Ωχ 1 h 2 more sharply constrains the SUGRA parameter space and the allowed band of event rates shrinks. The allowed range of the neutralino mass is also more sharply constrained since the mχ 1 mass is constrained to lie below ≈ 90 GeV, as above this mass Ωh 2 exceeds the upper limit of Eq. (57). Since the neutralino mass and the gluino mass scale one also deduces an upper limit on the gluino mass of about 650 GeV. Thus a precise measurement of Ωχ 1 h 2 will put an interesting upper bound on the gluino mass which would be testable at accelerators such as the upgraded Tevatron and the LHC. In general the overall effect of the more stringent constraints on Ωχ 1 h 2 is to eliminate a part of the parameter space allowed by Eq. (1) and thus limit more severely the permissible domain of the event rates. Conclusions In this paper we have investigated the effects of non-universalities in the Higgs secotor and in the third generation squark sector on the low energy parameters. We have shown that the Higgs sector non-universalities and the nonuniversalities in the third generation squark sector are highly coupled and we have exhibited analytic forms of their effects on the parameters µ, m A and the third generation squark masses at low energy. Next we investigated the effects of these non-universalities on the event rates in the scattering of neutralinos by nuclei. Here we found that the minimum event rates in the neutralino mass region 30 GeV ≤ mχ 1 ≤ 65GeV can be increased or decreased by a factor of O(10) depending on the sign of the non-universality. In contrast, in the neutralino mass region mχ 1 ≥ 65 GeV the minimum and the maximum event rates are not appeciably affected. These results have important implications for dark matter detectors. Supergravity unified models with universal boundary conditions for soft SUSY breaking parameters give event rates which lie in a wide range R= O(1 − 10 −5 ) events/kg da Thus while the current generation of dark matter detectors can sample a part of the parameter space, one needs far more sensitive detectors (more sensitive by a factor of 10 3 ) to sample the entire parameter space of the model. However, inclusion of non-universalities shows that for certain signs of the non-universality, the minimum event rates can increase by a factor of 10-100 in the region mχ 1 ≤ 65GeV . These models can be completely tested in this mass region if the detector sensitivity could be enhanced by a factor of about 10 2 . Such an enhancement does not seem out of reach with new techniques currently being discussed [55]. Another interesting result seen was the appearance of large event rates, up to 10 event/kg da for the case when δ 3 = δ 4 = −1. This case looks very interesting from the view point of experimental detection of dark matter. We have also analysed the implications of more stringent constraints on Ωh 2 which is expected from the next round of satellite experiments, and found that while the event rates typically become smaller a significant part of the parameter space yields event rates which are within reach of dark matter experiments currently being planned. Acknowledgments This research was supported in part by NSF grant numbers PHY-96020274 and PHY-9411543. Appendix A All the scalar sparticle masses receive contributions from the Tr(Ym 2 ) term. Corrections to the first two generations of sparticle masses are given by where S 0 and p are defined in the text. Under the assumption that the squarks and slepton masses are universal at the GUT scale for the first two generations, but with non-universaliites in the Higgs sector and in the third generation we can deduce the following sum rule for the Tr(Ym 2 ) term: Here T r(Y m 2 ) 0 is the value of T r(Y m 2 ) at the GUT scale while the right hand side is evaluated at the electro-weak scale. We see from above that if the masses of the first two generations of squarks and sleptons are known with enough precision, one can determine the value of the trace anomly term at the GUT scale. For the case when T r(Y m 2 ) 0 vanishes the above sum rule reduces to the one given in Ref. (56). m 2 H 2 , m 2 U and m 2Q satisfy the following set of coupled equations , and γ 1 =-3 10 , γ 2 = 2 5 and γ 3 =-1 10 and S is the running parameter with S(0)=S 0 . We show now that the evolution of the trace anomaly term is not affected by the top Yukawa coupling. To see this it is best to go to a decoupled set of equations using the functions e i (t) given by It is easily seen that the evolution of e 2 , and e 3 does not involve the Yukawa coupling and only the evolution of e 1 does. One has where and where m 2Q is defined in the text. Table 4: A comparison of the minimum and maximum event rates/kg da for xenon for the cases δ 1 = 0 = δ 2 , δ 1 = 1 = −δ 2 , and δ 1 = −1 = −δ 2 , and µ > 0 as a function of A t .	2018-12-15T04:41:35.879Z	1997-01-15T00:00:00.000	{ "year": 1997, "sha1": "cd342ea9ed17d95d586a53659d1497bfbc041ece", "oa_license": null, "oa_url": "http://arxiv.org/pdf/hep-ph/9701301", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "cd342ea9ed17d95d586a53659d1497bfbc041ece", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
229227181	pes2o/s2orc	v3-fos-license	Implementing Authentic Assessment in EFL Classroom to Prepare Pre-Service Teachers for Curriculum 2013 Artikel ini membahasa tentang Penerapan Penilaian Otentik di kelas EFL untuk Mempersiapkan Calon Guru dalam Kurikulum 2013. Penilaian Otentik adalah unit yang akan menjadi titik diskusi dalam artikel ini. Bagaimana Penilaian Otentik memiliki pengaruh besar untuk pemahaman dan persepsi calon guru. Penilaian Otentik adalah hal penting yang harus dipahami oleh calon guru. Penelitian ini melibatkan 10 responden yang merupakan mahasiswa semester 5 Program Pendidikan Bahasa Inggris di Universitas Ibn Khaldun. Tujuan dari penelitian ini adalah untuk membantu calon guru memahami dan mempersiapkan tentang Penilaian Otentik. Untuk mengumpulkan data, peneliti mengikuti beberapa tahap, yaitu mengisi kuesioner, wawancara, dan observasi. Dari metode kualitatif, peneliti melaporkan bahwa tanggapan responden sangat beragam. Temuan penelitian menunjukkan bahwa kebanyakan mereka tidak cukup memahami apa itu penilaian autentik. A. Introduction Authentic Assessment is multiple forms of assessment that reflect on student learning, achievement, motivation, and attitude on classroom activity (O'Malley & Pierce, 1996). According to the Ministry of Education and Culture No. 81, the year 2013 published that authentic assessment is an assessment which is focused on measuring student's learning process with their behaviour, knowledge, and skill. Accordingly, teachers should be required to assess student's skill using performance, project and portfolio assessment. Performance assessment is an assessment which is conducted by observing student's activity; portfolio assessment is a continuous assessment process based of a set of information that a given period time; while project assessment is an integrated unit which cannot be finished at a given time; it requires the students to do a step of tasks to collect the data (Minister of Education and Culture No. 104 the year 2014). Moreover, Authentic assessment replicates real-world challenges and standards of performance that skills typically by students (Koh, 2017). The implementation of the authentic assessment has some challenges for teachers, so the Minister of Education and Culture published curriculum 2013 to improve the implementation previous curriculum, which is the curriculum 2013 has relevancy with the authentic assessment. About the regulation of to Minister of Education and Culture No. 81a year 2013 about the implementation of curriculum 2013, teachers should be implemented authentic assessment as a student competence method. There are some steps in implementing a portfolio assessment in classroom activities. Those steps are: 1) the teacher should explain to the student that portfolio will give some benefits to teacher and students; 2) together with students, the teacher decides the sample of tasks; 3) the tasks are collected and organized into a special folder; 4) every task is identified based on the date, so the teacher can track the tasks of student progress during a given time; 5) together with student decides the criteria of scoring; 6) teacher can ask the student to check their work and at the same time help them to increased their skill; 7) if the students get a low score, teacher may give them second chances to improve with a given time; 8) finally, each of students work is collected into special folder (Gotlieb,1995). At some cases, in reality, the implementation of Authentic Assessment is not easy, because of the lack of teachers' knowledge about the curriculum 2013. Thought that the implementation of authentic assessment did not run effectively because of the complex procedure and the class condition. For the fluent of implementing authentic assessment should there a good collaborator between the teacher and students. Kurebwa dan Nyaruwata (2013) claimed that the problem of a teacher experienced on the implementing authentic assessment is the lack of teacher knowledge on assessment system, also teacher more used the summative assessment than formative assessment. Furthermore, in previous, this problem was researched by Retnawati at 2016 with the title Vocational High School Teachers' Difficulties in Implementing the Assessment in Curriculum 2013. The lack of training in previous for teacher become a reason was not easily the authentic assessment to (Retnawati, 2016). There are many reasons, such as 1) the obstacle in-class activities; 2) the lack of themselves knowledge; 3) the parent involvement low; 4) the string to the norm. Moreover, Bonnie Keilty (2016) with the title Early Interventionists' Reports of Authentic Assessment Methods Through Focus Group Research described that in the implementation of authentic assessment should be focused to attitude and requirement the professionals to implement that. But, right now still not clear how an intervention which used clearly in authentic assessment what has an equal with curriculum 2013. Requires a maximal effort from implementing and or realization of authentic assessment to increase what is assessment expected. This concern has the aim to know the perception and the knowledge of an amateur teacher with the implementation of authentic assessment, and what is this implement can increase the knowledge of an amateur teacher in the future. B. Research Method The research design was a descriptive study where the researcher used quantitative and qualitative method. The data collection used 3 steps, which are observation, questionnaire, and interview. The setting was at 6 December 2019a in around FKIP Universitas Ibn Khaldun. According to Richard Tewksbury (2009), the qualitative method is a research which is focusing on a value and a knowledge of how someone understood something. Moreover, Gumilar Rusliwa (2005) described that the qualitative method explains that the qualitative method more focusing on documenting and process which related to managing a character from the daily activity in reality. The First step is giving a questionnaire to 10 respondents which from class 5K English Education Department, Universitas Ibn Khaldun. The questionnaire focused on a statement related to this study for getting some relevant information with this study. The instrument consists of 10 questions with for Likert scale was also chosen. The next step was we did to interview with 10 respondents. Structured Interviews is a kind of interview which is used in this study. According to Easwaramoorthy & Fataneh (2006) explained that Structured Interviews is a kind of interview which is a question have a set from the researcher and equals with a relevant topic. The last step was observation for collecting data which we get from our interview and questionnaire. From the last step, we could get the relevant data to help us make a summary. C. Results And Discussion The observation is the last steps which we did to collecting data. Before we did the interview step, we should to ask the respondents to fill the questionnaire which is we made before and an equal with this study. Statement 1 Statement 2 The data from the percentage of the first statement shows that most of the respondents involved do not know what Authentic Assessment is. However, not a few of the respondents who already know the meaning of Authentic Assessment, but still cannot understand how the application must be done. Another data shows a percentage of the second statement. The respondents who asked to give an example of an authentic assessment, 50% of them could not. They cannot mention it at all. Statements 3 In addition to the questionnaire statement that discusses the definition and examples of Authentic Assessment, we also provide a statement that journal articles can help improve understanding in digital literacy. Can be seen in terms of the percentage above, 67% of respondents agreed to it. Also E-ISSN : 2614-4271 P-ISSN : 2599-1302 supported by the reason of the interview results of one of the respondents namely "with a reading a journal article, I can add my knowledge, and it can be a reference when we will make an article", said the 3rd respondent. Statements 4 It can also be seen from the percentage of statement 4 above that 70% of respondents agreed that reading and understanding articles could increase respondents' understanding of education. One of the respondents we asked, the 5th respondent said "many methods can I get and then I can apply at my studies" The percentage of statement 5 shows, 34% of the respondents agreed that the existence of assignments related to the article was able to help them in terms of adding their insights and knowledge as stated by the 6th respondent "My insight is increasing about the article. Now I can find out various types of articles and journals, and can help me in applying Authentic Assessment in C13 ". It shows that this assessment can help the respondents in understanding an Authentic Assessment through articles. Looking up to at all of the data taken, explaining that the application of Authentic Assessment is not something easy to do, especially to be applied in the Curriculum 2013. Prospective teachers who are prepared to act on the implementation of Authentic Assessment must be truly supported, given understanding and training so that the competency of teacher candidates can improve and implement their objectives well. D. Conclusions A teachers' understanding of authentic assessment is more be references for whether the authentic assessment can be applied. Furthermore, in the 2013 curriculum is something that a teacher must be obliged to support his performance through Authentic Assessment. Although it is not easy, prospective teachers must understand and understand the intended purpose. The point is to prepare the teacher candidates for actual practice. The 2013 curriculum is very closely related to an Authentic Assessment, and therefore the role of the teacher is needed to help implement the program well. Good collaboration between students and teachers can help this assessment run as it should.	2020-11-19T09:13:30.811Z	2019-12-31T00:00:00.000	{ "year": 2019, "sha1": "2e88eeff12c432e8d2d9be55b2d838483ca2633b", "oa_license": "CCBYSA", "oa_url": "https://doi.org/10.24090/tarling.v3i1.3481", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "97fb0f14484e8777cca538234b1f0539fe78f626", "s2fieldsofstudy": [ "Education" ], "extfieldsofstudy": [ "Psychology" ] }
229462189	pes2o/s2orc	v3-fos-license	Development of rapid heat solid solution process of Al 7075-T6 alloy applying near-infrared ray heater To improve the formability and productivity in cold forming of aluminum alloy, w-temper forming process was developed. In this process, the T6-temper blank is subjected to the solution heat treatment (SHT) and then quenched to achieve the w-temper. At present, the main problem of this process is that too long SHT time cannot meet the manufacturing cost. For example, the quenching time is 10 sec, while the SHT time is 30 min. In this paper, rapid solid solution of Al 7075-T6 alloy sheet was implemented in near-infrared ray(IR) heater. Applying IR, contact heater was developed instead of the traditional heater using cartridge heater. The results have shown that it is possible to accurately control the sheet temperature for solid solution temperature in short time. Also, compared to SHT using traditional heater, the rapidly heating using IR heater could make the mechanical properties of w-temper. Introduction Global warming is a serious concern where it is driven in part by the exhaust gases from automotive vehicles. For both fuel economy and emissions control, reducing the weight of car bodies has been a focused issue in the automotive industry [1][2][3]. For weight reduction, Al7xxx alloys have attracted much attention from the automotive industry due to its superior specific strength compared to ultrahighstrength steel (UHSS). But, a disadvantage with high-strength these alloys compared with UHSS is that the formability is low. In order to improve the formability and productivity in the cold forming of these alloys, a limited study has suggested the application of W-temper forming (WF). In WF process, the T6-temper blank is subjected to the SHT and then quenched to achieve the w-temper. After that, the blank is transferred into the die and deformed at RT before the precipitation starts. At present, the main problem of this process is that too long solid solution time cannot meet the manufacturing cost. Several studies have demonstrated that the rapid heating of aluminum alloys is significantly influenced by short heating time. In practice, Xu, Xiaofeng et al. examined that heating speed by contact solid solution treatment show a critical influence on microstructure and mechanical properties of Al 7075-T6 alloy [4]. Maeno, Tomoyoshi et al. demonstrated that quick heating provides similar mechanical propertied and formability [5]. In this paper, the innovative heating methods in a contact IR heating was utilized to reduce the solid solution time of Al 7075 alloy sheets. First, a contact IR heater was designed and manufactured. And then the research on the rapid SHT of Al 7075 alloy by contact heating method was conducted. Experiment A commercial Al7075 alloy in T6 condition with a thickness of 2.0 mm has been used in this research, having the following chemical composition given in table 1. In order to simulate the WF process, Al 7075 alloy samples were solution heat treated in a contact IR heater for 150 sec at 475°C. Then, all the samples were quenched by a cold flat die. The WF process is schematically shown in Figure 1. Hardness test was conducted to observe the change in mechanical property in the alloy with different solid solution time. The measurements were operated using a HM-100 Vickers microhardness tester with a load of 100 gf and a dwell time of 12 sec. Thermal analysis was carried out using the differential scanning calorimetry (DSC) system (TA-100, TA Instruments). High purity Al 99.9% was used as a reference material. The specimens were heated from 60 to 300 °C at a constant heating rate of 10 °C min -1 . The selected conditions of samples were prepared for the DSC testing to have thicknesses of ~0.6 mm with weights of ~20 mg. Contact IR heater setup IR heating has more advantages than traditional heating. However, due to the temperature accuracy of sheets during heating, there are some requirements for the design of contact IR heater such as excellent temperature uniformity, minimize overshoot with temperature controllers and heating time. On the basis of important points, a contact IR heater was designed, as shown in Figure 2(a). It consists of the upper IR heater and lower cartridge heaters. The upper was designed to rapidly heat on the sheets by the output power of the IR heater controlled by the PID control system (Figure 2(b)). The lower die was selected for copper because of its high thermal conductivity and designed to maintain the temperature uniformity of the sheets during the control IR heater by cartridge heaters. Spring loaded K-type thermocouple were installed inside the lower die between each of cartridge heaters feedback adjustment of temperature uniformity and minimizing overshoot of the sheets. Guide pins were arranged in the lower die to assist the placement of the sheets (Figure 2(c)). Finally, the contact IR heater did not need to be installed in the press and equipment was driven to open and close by the movement of the pneumatic cylinder. The sheets were placed on the lower die, and the heating is completed after the equipment is closed, as shown in Figure 2 Measurement of temperature The temperature evolution of the sample at 475 °C with a holding time of 150 sec is shown in Figure 3. The temperature-time curve can be divided into two stage. In the first stage, when the sample temperature increased to 475 °C , the heating rate was very fast, and the sample reached 475 °C within 90 sec. In the second stage, when the samples reach target temperature, under the controlling of PID control system, the temperature fluctuates up and down at 475°C and gradually stabilizes. The temperature of the sample is relatively uniform and the temperature difference is about 2 °C . . Result of Hardness Before forming at room temperature, the samples require sufficient solid solution time. In order to study the effect of solid solution time, the hardness of the w-temper with different solid solution time was measured. Figure 4 shows the measurement of hardness for four different times, 90, 105, 120 and 150 sec. It can be seen that hardness has decreased compared with the as-received condition (T6 ≈ 178Hv). Also, all samples show the hardness of Hv ≈ 88 immediately after quenching. The decreased hardness with different solid solution times is attributed to microstructure immediately after quenching. It is implied that the dissolution of MgZn2 precipitation occurred. Result of DSC Significance of solid solution time on the precipitation behavior was examined through the DSC analysis for WF capability on the Al 7075 alloy. It is well described that the precipitation sequence of Al 7xxx alloys is expressed as follows [6,7]: Super saturated Solid Solution → G.P. Zones (GPZs) → metastable η' phase → stable η phase (MgZn2). The DSC curves of the WF samples measured at four different times, 90, 105, 120 and 150 sec after SHT are given in Figure 5. It was reported earlier through the DSC examinations that the first peak reaction (stage I) at ~200 °C corresponds to the formation of the GPZs, the following double exothermic reactions (stages II and III) at ~210 and ~230 °C are attributed to the formation of the η′ phase and the transformation of η′ into the η phase. All samples demonstrated that first peak reaction appears at approximately 95 °C, most likely due to the formation of GPZs. The GPZs peak could be detected at DSC curve, which means the soluble Al matrix were completely dissolved.	2020-11-26T09:03:52.209Z	2020-11-19T00:00:00.000	{ "year": 2020, "sha1": "ac309a83bf8a56e3cbbe97f312e2f03423ec95bf", "oa_license": null, "oa_url": "https://doi.org/10.1088/1757-899x/967/1/012070", "oa_status": "GOLD", "pdf_src": "IOP", "pdf_hash": "6f5133a55d4719f44575b4f533cf8c9c6ee0d6c0", "s2fieldsofstudy": [ "Materials Science" ], "extfieldsofstudy": [ "Physics", "Materials Science" ] }
261932655	pes2o/s2orc	v3-fos-license	Basic thermodynamic description of adsorption of polar and nonpolar molecules on AOGW . In this article, the differential thermal adsorption, adsorption isotherm, integral entropy, and thermokinetics results and their analysis of the polar molecule H 2 O vapor selected as adsorbate in the adsorbents obtained by the traditional method of thermal and water vapor activation of grape seed waste under the influence of temperature and their analysis, Q d , ΔF, ΔS and isotherm, curves of differential heat, and kinetics of the experimental research results are presented, as a result of which the law of adsorption in active canter’s is determined. Introduction Today, it can be seen from the literature analysis that adsorbents are obtained from natural mineral compounds, mainly clay minerals, and other sources of raw materials.Adsorption of adsorbents and zeolites, including the adsorption of polar molecules on bentonites as well as the adsorption of organic and inorganic substances on synthetic zeolites, is known to us from the research works of the authors [1][2].So far, in order to study the changes in the adsorption volume of adsorbents activated on the basis of tree trunks and their thermodynamic characteristics, research on the mechanisms of adsorption of ascorbates on adsorbents by the pyrolysis method or activation using water vapor at different temperatures has been studied [3][4][5]. To date, complete thermodynamic descriptions of adsorbents obtained on the basis of waste, taken precisely on the basis of grape compression, and data on the mechanisms of adsorbents Adsorbents with adsorbents have not been studied [6][7][8][9][10].The results of Differential heat adsorption, H2O vapor adsorption, integral entropy, thermokinetics, and their analysis were all studied in adsorbents that were made by activating water vapor with heat and water vapor under a certain pressure [11,12].By looking at these data on the adsorption of activated adsorbents in adsorbates and the adsorption of adsorbents forming a complex in coordination spheres, we can learn important basic information about where adsorbents can be used and how much adsorbate is adsorbed.[13][14].Specific properties of adsorbent obtained from grape juice using water vapor (AOGW) type adsorbents, nature of hydrophobicity, thermodynamic processes taking place on the adsorbent surface with polar water vapor molecules-adsorbate, mainly surface chemistry and adsorption of solid phases with pore structure and active centers, provide important information for their use.Based on the determination of thermodynamic parameters of adsorbents of different natures in the calorimeter method as well as the analysis of the areas of absorption going into their active centers, it is possible to find out the complete data of the adsorbent and catalyst adsorption mechanisms [15][16][17][18][19]. Today, thermodynamic characteristics of adsorbents provide data on their adsorption capacity; it is known to have data on the capture of such important results as isotherm, differential heat, entropy, and thermal equilibrium time of water vapor adsorption [12,[20][21][22][23][24][25][26][27][28][29][30][31]. Research Methods This article uses adsorption-microcalorimetric methods of analysis.In determining the thermodynamic parameters of adsorbents with high adsorption activity, the process of studying the adsorption mechanism used a universal high-vacuum adsorption device at a temperature of 303 K and a system of differential microcalorimeter-type differential temperature adsorption meters and a precision high-sensitivity high-vacuum microcalorimeter Tiana-Calve DAK-1-1 connected to it.Using the device, the values of the isotherm, differential heat content, entropy, and kinetics of water vapor adsorption on the adsorbent (thermal adsorbent derived from grape juice (TADG-6) and adsorbent obtained from grape juice using water vapor (AOGW-6)) were obtained. Results and Discussion The results of the study carried out on the adsorbents TADG-6 and AOGW-6, which were selected as the most optimal of the adsorbents that were activated using waste, are presented.The results of the experimental study showed that at a temperature of 303 K in a device with a microcalorimeter, when an H2O molecule was selected as a polar molecule for adsorption in the adsorbent AOGW-6, a complete thermodynamic process was carried out.On the basis of the results of adsorption, entropy, and thermokinetics of H2O vapor molecules in the adsorbent AOGW, activated with the participation of water vapor on the basis of grape seed waste, the law of adsorption in active centers was established on the basis of the differences in the adsorption mechanisms of these adsorbents.Figure 1 shows the results of the isotherm of adsorption of the polar molecule H2O in the activated adsorbent AOGW-6 using water vapor.The results of the experimental study of the water vapor adsorption isotherm in the activated AOGW-6 adsorbent are shown to be able to be fully described by the two-member equation of the theory of volumetric filling of micropores (TVFM).Figure 1 shows the results of the isotherm of the adsorption of the molecule H2O in the adsorbent AOGW-6.These results go along with the results of the differential heat of adsorption and help draw a clear conclusion about the full thermodynamic characteristic of the adsorbent, which is the adsorption mechanism. Fig. 2. Heat of adsorption of H2O vapor molecules in activated adsorbent The adsorption heat curvature of the H2O vapor molecules in the adsorbent AOGW-6 activated using water vapor is shown in Figure 2, where it can be observed that the adsorbent adsorption of a polar water molecule, selected as an adsorbate, is stepwise adsorbed in the adsorbent.On the graph, the amount of curvature (Qd) is shown in kJ/mol, and the amount of adsorption (a) is shown in mkmol/g.Initially, differential issyclic adsorption Qd was observed to adsorb at the highest value of 57.81 kJ/mol and an adsorption rate equal to a=6.36 mkmol/g, indicating that this higher value is adsorbed in existing adsorbent-containing additives.Later, the value of differential heat adsorption was characterized by the fact that adsorption issicity variation up to Qd= 48.07 kJ/mol was observed with adsorption up to a=44.64 mkmol/g.The initial I-coordination sphere begins with the value of Qd= 48.07 kj/mol of the differential adsorption heat and a=44.64 mkmol/g of the adsorption quantity, each of the three water molecules adsorbed uniformly from a quantity of 110 mkmol/g, a process that continues until the value of the adsorption heat Qd= 45.20 kj/mol and the In this case, if each water molecule forms a CI:H2O complex (Qd= 48.07÷46.10kj/mol and a=44.64÷156.7 mkmol/g) in the first adsorption in this coordination sphere, in the next step 2, 2si:adsorption of the H2O complex (Qd= 44.64÷46.4kj/mol and a=156.7÷270.64 mkmol/g) based on the adsorbent-adsorbate mechanism, and in Step 3, with the same amount, the 3CI:H2O complex (Qd= 46.4.45.20 kdj/mol and a=270.64÷356.66mkmol/g) in the formation of differential heat there was a change in adsorption, varying around 4 kj/mol.In the next adsorption, however, adsorption continues in the II coordination sphere, in which the adsorption of C:H2O by forming a complex (Qd= 45.20÷44.90kj/mol and a=365.66÷443.67 mkmol/g) continues.Further adsorption in the active center belonging to this coordination sphere (Qd=44,90÷45,00 kj/mol and a=443,67÷521,35 mkmol/g) leads to the complex formation of 2C:H2O.It was observed that, based on the analysis of the results of the differential heat of water vapor adsorption of the AOGW-6 pargase activated adsorbent, the differential heat of adsorption of the H2O molecule in the AOGW-6 adsorbent (Qd) was observed to adsorb in the adsorbent in the proportion of 110 mkmol/gga and 80 mkmol/gga in active centers.The results of the differential adsorption entropy of water vapor molecules in the activated adsorbent AOGW-6 are shown in Figure 3. Based on the results of this study, it can be concluded that the fact that integral enropia is found around about 14 kJ/molK means that adsorption is close to the solid state.The molly differential entropy (ΔSd) of H2O adsorption in the AOGW-6 adsorbent was found, calculated from the isotherma and differential heat of adsorption based on the Gibbs-Helmgols equation.In the AOGW-6 adsorbent, the water entropy is the molly differential entropy of ΔSd adsorption, which is different from its liquid state water entropy and has a waveform appearance.Starting from 38.46 J/molK, ΔSd increases to ~30.7 mmol/g with small saturations, showing a relatively high saturation and crossing the maximum.The adsorbentadsorbent interaction ensures this situation.At AOGW-6, the entropy of water ΔSd passes through the lowest minimum (-7.75 j/mol), which is different from the entropy of solid water, while towards saturation it rises sharply to the values of the Liquid Entropy of water.The integral mean moly entropy is equal to -16.24 J/molK, which is much lower than that of Liquid Entropy.Thus, differential and integral entropy in the adsorbent AOGW-6 showed strong braking, in which the entropy of water is close to its entropy in its solid state.Analysis of the results of the adsorption thermokinetics of water vapor molecules in the adsorbent AOGW-6 (fig.Based on the data of adsorption Kinetics, it can be said that adsorption first went for 6 hours and 30 minutes and then went for around 2 hours in the last parts of adsorption, indicating the stability of the adsorbate in the adsorbent.The results of the study, which are presented, are based on the results of the comparative differential heat of H2O vapor adsorption in the carbon-based adsorbent AOGW-6, the adsorption heat of H2O vapor molecules activated in the presence of water vapor based on isotherma, entropy, and thermokinetics, as well as the waste of Grape Grain.It was found that in studies using high-vacuum micrakolorimetric research methods, the differential heat of adsorption of the H2O molecule in the water vapor-activated AOGW-6 adsorbent (Qd) results in an adsorbent of 110 mkmol/g in microgroups and an adsorbent-adsorbate interaction mechanism in mesagovaks as a result of adsorption in 80 mkmol/g carrally active centers.Based on the complete thermodynamic characteristics (Qd, ΔF, ΔS) and isotherm of thermally and pargase-activated TADG-6 and USOP-A-6 adsorbents (Qd, ΔF, ΔS), the correlation law of CO2 adsorption was determined in the active centers of the adsorbent at a rate of 500 mkmol/g.According to the results of the differential heat of adsorption of the H2O molecule (Qd) in the USOP-A-6 adsorbent, the adsorbent-adsorbate interaction mechanism as a result of adsorption in adsorbent microgroups by 110 mkmol/g and in mesagovaks by 80 mkmol/g in multiples In the course of the experiments, complete thermodynamic descriptions of the adsorbents TADG-6 and AOGW-6, which were activated using thermal and water vapor, i.e., based on Qd, ΔF, ΔS and isotherma, from the results of an experimental study, are presented, as a result of which the Conclusion The results of the study were carried out on the basis of carbon-based AOGW-6 adsorbent H2O vapor adsorption, isotherma, entropy, and thermokinetics of AOGW-6 adsorbent activated in the presence of water vapor based on Grape Grain waste, as well as the results of adsorption heat, isotherma, entropy, and thermokinetics of the information based on it. Fig. 3 : Fig. 3: Adsorption entropy of water vapor molecules in the adsorbent AOGW-6In the AOGW-6 adsorbent, the water entropy is the molly differential entropy of ΔSd adsorption, which is different from its liquid state water entropy and has a waveform appearance.Starting from 38.46 J/molK, ΔSd increases to ~30.7 mmol/g with small saturations, showing a relatively high saturation and crossing the maximum.The adsorbentadsorbent interaction ensures this situation.At AOGW-6, the entropy of water ΔSd passes	2023-09-16T15:06:34.423Z	2023-01-01T00:00:00.000	{ "year": 2023, "sha1": "88467d005cd3462c1790e485a10f726ab5a27f8c", "oa_license": "CCBY", "oa_url": "https://www.e3s-conferences.org/articles/e3sconf/pdf/2023/62/e3sconf_ictced2023_04003.pdf", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "fd493d295884a4d2fa5a107c599d6178385a1434", "s2fieldsofstudy": [ "Chemistry", "Environmental Science", "Materials Science" ], "extfieldsofstudy": [] }
254902480	pes2o/s2orc	v3-fos-license	Predictive Ability of the Three-Time Stand and Walk Test to Determine Frailty and its Associations with Fear of Falling and Cognitive Function in Community-Dwelling Older Adults Background The three-time stand and walk test (TTSW) is a complex functional task used to determine muscle strength, balance, and fall risk in older individuals. This study hypothesized that TTSW is an appropriate tool for detecting frailty related to falls and the cognition of community-dwelling older adults. The study objectives were to assess the ability of the TTSW to determine frailty by exploring the optimal cut-off score, and to investigate the correlations between TTSW outcomes with falls and cognitive function in 118 community-dwelling older adults. Methods The demographic data of eligible participants were assessed, and the participants were diagnosed with frailty based on the frailty phenotype. The participants then completed the Frail Non-Disabled (FiND) questionnaire, TTSW, Falls Efficacy Scale-International (FES-I), and Mini-Mental State Examination Thai version (MMSE-Thai 2002). Results The results demonstrated that the TTSW outcomes were significantly correlated with FiND, FES-I, and MMSE-Thai 2002 (rho=0.705, r=0.482, and r=-0.510, respectively; p<0.001). Moreover, a TTSW time of 18 s or longer had a good ability to indicate frailty in older individuals (sensitivity=88.41%, specificity=83.67%, area under the receiver operating characteristic curve [AUC]=0.926). Conclusions Implementing this tool in a community setting may be useful for the initial screening, monitoring, and referral of data by healthcare professionals. A cut-off TTSW time of 18 seconds or longer was the optimal criterion to indicate frailty in community-dwelling older people. phenotype approach, or Fried's frailty phenotype, classifies an individual as frail if they present three or more of the five frailty items-slow walking speed, impaired grip strength, declining physical activity levels, exhaustion, and unintended weight loss. The presence of one or two of the five items is defined as pre-frailty, while individuals with none of these items are considered healthy or robust. 11) In addition, many screening tools are available to assess the risk of frailty. However, most of these are in the form of questionnaires 11,13) for qualitative assessment, including the Frail Non-Disabled (FiND) questionnaire, Clinical Frailty Scale, and simple frailty questionnaire (FRAIL). 14,15) These are commonly applied through interviews and observations; thus, the outcomes are subjective and highly dependent on the experience of the assessors. 16) In addition, decisions about clinically relevant changes can be difficult to make when the time intervals between visits are long or when assessors change. 17) By contrast, functional assessments are objective and roughly indicate the functioning levels of individuals; moreover, the outcomes may suggest further assessments for systems with disorders. [18][19][20] Therefore, functional assessments, as assessed by Fried's frailty phenotype, may be more useful and are commonly used as screening and monitoring tools over time. 20) However, this standard method of assessment is based on a combination of multiple tests, which can be time-consuming and may be inconvenient for use in communities with large populations. Therefore, the exploration of a simple functional test reflecting the overall functional ability that contributes to frailty is crucial and may promote the effectiveness of prevention and management strategies for older individuals. Frailty is often determined by important physical abilities, including strength, gait ability, and balance. Therefore, we previously developed a three-time stand and walk test (TTSW) to assess muscle strength, gait, and balance ability in community-dwelling older individuals. 21) The test requires individuals to perform sit-tostand movements on a chair three times, walk around a traffic cone placed 3 m from the front edge of the chair, and return to sit down on the chair at the fastest safe speed. 21) The test has excellent reliability (intraclass correlation coefficient [ICC] = 0.991; 95% confidence interval [CI], 0.984-0.996) and good correlation with standard tests, including the five-time sit-to-stand (FTSST) (r = 0.648, p < 0.001) and Timed Up and Go test (TUGT) (r = 0.673, p < 0.100). 22) Moreover, the outcome of TTSW was used to indicate the risk of falling in older individuals. 21) Therefore, the present study hypothesized that the TTSW is an appropriate method for detecting frailty related to falls and cognition in community-dwelling older adults. The objectives of this study were to assess the predictive validity (primary objective) by exploring the optimal cutoff score of the TTSW and to determine frailty and concurrent va-lidity (secondary objective) by investigating the correlation between TTSW outcomes with falls and cognitive function in community-dwelling older adults. MATERIALS AND METHODS The cross-sectional study involved 118 Thai community-dwelling older people aged 60 years and above and a body mass index (BMI) between 18.5 and 29.9 kg/m 2 . For eligibility, the participants were required to be able to stand up independently, walk with or without a walking device, and understand the commands used in this study. The exclusion criteria were any signs and symptoms that might affect study participation, such as uncontrolled or unstable hypertension, heart disease, and pain in the musculoskeletal system with a visual analog scale score of more than 5 out of 10. The research protocols were approved by the Ethics Committee of the Institute for Human Research (No. 2/064/61), University of Phayao, Thailand. The participants provided written informed consent before their participation in the study. This study complied the ethical guidelines for authorship and publishing in the Annals of Geriatric Medicine and Research. 23) Procedures The eligible participants were assessed to collect demographic data, including age, sex, height, weight, vital signs, underlying diseases, requirement for a walking device, and fall history over the past 6 months. The participants were diagnosed with frailty based on the frailty phenotype criteria. 11) The participants were then assessed for the study outcomes ( Fig. 1), including the following. FiND questionnaire This study used the FiND questionnaire as a standard screening tool to determine frailty. The FiND questionnaire consists of five questions. Two questions are specifically aimed at identifying individuals with mobility disabilities (an early stage of the disabling process). For the present analysis, we defined the presence of mobility disability as "a lot of difficulties" or "inability" to perform at least one of these two terms. The remaining three questions are aimed at assessing the following signs, symptoms, or conditions that are commonly considered components of frailty syndrome: weight loss, exhaustion, and sedentary behavior. In the present analyses, we considered participants meeting one or more frailty criteria in the absence of mobility disability to be "frail. " Notably, the weight loss and exhaustion criteria included in the FiND questionnaire were the same as those originally proposed for the frailty phenotype. Participants reporting no mobility disability and no frailty criteria were considered robust in the FiND questionnaire. 11,14,24) TTSW The participants were required to sit on a standard armless chair with their backs upright, a hip flexion of 90°, feet placed flat on the floor approximately 10 cm behind the knees, and arms at their sides. The participants were then instructed to stand with their hips and knees in full extension and then sit down three times, walk around a traffic cone placed 3 m from the front edge of the chair, and return to sit down on the chair at the fastest safe speed. 21) The average time to complete the three trials was used in the data analyses. During the tests, the participants wore a lightweight safety belt around their waist, with a physiotherapist walking alongside them to ensure the safety of the participants and improve the accuracy of the outcomes. Fear of falls (FOF) assessment We tested the participants' FOF using a yes/no question evaluation scale combined with the Falls Efficacy Scale-International (FES-I). 25) The FES-I was developed and validated for the intensive assessment of FOF, self-efficacy, and balance confidence. The questionnaire comprises 16 items, including basic and instrumental activities of daily living. Each activity was scored from 1 (not at all concerned) to 4 (very concerned), resulting in a total score ranging from 16 (absence of concern) to 64 (extreme concern). Mini-Mental State Examination (MMSE) Thai 2002 This study used the Thai version of the MMSE (MMSE-Thai 2002). The test was developed and validated using the original English version. 26) The MMSE-Thai 2002 is a screening instrument commonly used as a global cognitive test and is the current clinical mainstay cognitive screening instrument in Thailand. The test was scored in terms of the number of correctly completed items, in which low scores indicated poor performance and significant cognitive impairment. This 30-item questionnaire (the maximum score is 30) is used extensively in clinical, research, and community settings to measure and screen for cognitive impairment. 27,28) Statistical Analysis We used descriptive statistics to explain the personal data and study findings. An independent sample t-test was used to indicate the discriminative ability of TTSW for participants with and without frailty. Spearman rank correlation coefficient (rho) and Pearson correlation coefficient (r) were used to quantify the levels of correlation between the outcomes of TTSW and those of FiND, FES-I, and MMSE-Thai 2002 scores. Receiver operating characteristic (ROC) curves were further utilized to verify the diagnostic accuracy of the TTSW to indicate frailty by exploring the optimal cut-off score, sensitivity, specificity, and area under the ROC curve (AUC). The level of statistical significance was set at p < 0.05. RESULTS The current study recruited a total of 118 eligible participants with an average age of 73 years and a normal BMI. The average age and BMIs differed significantly between frail and normal participants (p < 0.001 and p < 0.05, respectively) ( Table 1). Although most participants could walk independently without assistive walking devices (77.97%), all those who used walking devices (26 participants) were frail. The results showed significant differences in the TTSW, FES-I, and MMSE-Thai 2002 scores between participants with and without frailty (p < 0.001) ( Table 1). Fig. 2 is a scatterplot showing the levels of correlation between the outcomes of the TTSW and the FiND, FES-I, and MMSE-Thai 2002. The results revealed that the TTSW scores were significantly correlated with FiND, FES-I, and MMSE-Thai 2002 (rho = 0.705, r = 0.482, and r = -0.510, respectively; p < 0.001) (Fig. 2). Table 2 shows the optimal cut-off scores for the TTSW. The results indicated that a TTSW time ≥ 18 seconds had a good ability to indicate frailty in older individuals, with the best sensitivity, specificity, and accuracy. DISCUSSION The results of the present study verified the TTSW as an alternative optimal screening tool to detect frailty in community-dwelling www.e-agmr.org older adults. The key finding of the current study was that the TTSW was significantly correlated with frailty, FOF, and cognitive function, as measured using FiND, FES-I, and MMSE-Thai 2002, respectively. Moreover, the results of the current study showed that the time to complete TTSW was at least 18 seconds, which was the optimal cut-off score to indicate frailty in community-dwelling older individuals. The TTSW was developed to assess the functional capacity and risk of falls in community-dwelling older individuals. 21) The test is challenging and demanding for muscles in the lower extremities and for assessing balance control and walking ability. This ability requires adequate muscle co-contraction, muscle force, and joint torque in the lower extremities to repeatedly perform sit-to-stand movements. Moreover, the ability to stand from a chair at the fastest safe speed three times can determine the functional capacity and risk of falls in community-dwelling older adults. 29) The ability to perform sit-to-stand movements three times was significantly correlated with FTSST (r = 0.942, p < 0.001) and was an excellent fall indicator (sensitivity = 88.89%, specificity = 100%, AUC = 0.92, 95% CI 0.81-1.00). 29) In addition, the participants were requested to walk at the fastest and safest speed for 3 m and then return to sit in the testing chair. This ability requires good postural control while walking at the maximum speed and turning. In particular, the ability to walk at a maximum and safe speed requires good neuromuscular function 30) and can be a valuable assessment tool to more rapidly determine functional decline than that based on usual walking speed with advancing age. 31,32) Moreover, the assessment of walking speed has shown validity in predicting frailty. 33,34) Therefore, the results of the test were significantly positively correlated with functional strength and balance in community-dwelling older people (r = 0.88, p < 0.001 and r = 0.91, p < 0.001, respectively) 35) and can be a good fall screening tool, with high sensitivity and specificity and a high AUC (sensitivity =80.00%, specificity =91.43%, AUC = 0.87, 95% CI 0.79-0.98). 21) Frailty leads to poor outcomes in older people. 11) In particular, frailty is associated with physical capacity impairments, such as muscular weakness and slow walking speed, 11) and decreases in other physical capacities, including reduced balance, speed, endurance, dexterity, and muscle density. 36,37) Older individuals who experience falls often develop FOF when performing daily activities. 38,39) The results of the present study demonstrated that falls are an independent predictor for developing FOF 20 months later (OR = 1.75; p < 0.001), and that FOF at baseline was a predictor of falling at 20 months (OR = 1.79; p < 0.001). 39) An intense FOF caused a 1.45-fold higher risk of cognitive decline in older adults compared to that in individuals without FOF. 40) In particular, the participants with and without frailty showed significant differences in executive functions and processing speed domain (p < 0.01), as measured by the MMSE. 40) This study has some limitations. First, many methods were used to screen for frailty in older individuals, which may have affected the identification of the participants in this study compared to other studies and their future use. In particular, the current study was conducted in older Thais; therefore, caution is needed when applying the results to other populations that may have different baseline characteristics. Second, two-thirds of the participants were women, which may have affected the results. However, when the participants were divided into frail and non-frail groups, the proportions of male and female participants were equal. Third, this study utilized the FiND to identify frailty; however, this tool may not be consistent with other international definitions of frailty. Further studies may need to apply gold-standard diagnostic criteria such as the Cardiovascular Health Study (CHS) frailty phenotypes. In conclusion, the results of the current study support the use of the TTSW to determine frailty in community-dwelling older adults. TTSW showed significant correlations with the outcomes of frailty, FOF, and cognitive function (rho = 0.705, r = 0.482 and r = -0.510, respectively; p < 0.001) (Fig. 2) and had excellent diagnostic accuracy in determining frailty with high sensitivity (88.41%), specificity (83.67%), and AUC (0.926; 95% CI 0.883-0.970) ( Table 2). The TTSW is a complex but practical and inexpensive tool with minimum equipment requirements. The implementation of this tool in a community setting may be useful for the initial screening, monitoring, and referral of data by healthcare www.e-agmr.org professionals. A TTSW cut-off score of 18 seconds was the opti-	2022-12-21T16:12:24.570Z	2022-12-01T00:00:00.000	{ "year": 2022, "sha1": "186e717675368e3bbb0688cf43316f39c6e13eae", "oa_license": "CCBYNC", "oa_url": "https://www.e-agmr.org/upload/pdf/agmr-22-0072.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "f49375e799415228b5fe498b8844ae9513b0bbd7", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
256690456	pes2o/s2orc	v3-fos-license	Boosting the electronic and catalytic properties of 2D semiconductors with supramolecular 2D hydrogen-bonded superlattices The electronic properties of two-dimensional semiconductors can be strongly modulated by interfacing them with atomically precise self-assembled molecular lattices, yielding hybrid van der Waals heterostructures (vdWHs). While proof-of-concepts exploited molecular assemblies held together by lateral unspecific van der Waals interactions, the use of 2D supramolecular networks relying on specific non-covalent forces is still unexplored. Herein, prototypical hydrogen-bonded 2D networks of cyanuric acid (CA) and melamine (M) are self-assembled onto MoS2 and WSe2 forming hybrid organic/inorganic vdWHs. The charge carrier density of monolayer MoS2 exhibits an exponential increase with the decreasing area occupied by the CA·M unit cell, in a cooperatively amplified process, reaching 2.7 × 1013 cm−2 and thereby demonstrating strong n-doping. When the 2D CA·M network is used as buffer layer, a stark enhancement in the catalytic activity of monolayer MoS2 for hydrogen evolution reactions is observed, outperforming the platinum (Pt) catalyst via gate modulation. Here, the authors report the functionalization of monolayer transition metal dichalcogenides with hydrogen-bonded 2D supramolecular networks of cyanuric acid and melamine, leading to a pronounced n-doping effect and enhancement of MoS2 catalytic activity for hydrogen evolution reactions. S ince the first isolation of graphene 1 , the family of twodimensional materials (2DMs) has expanded rapidly providing access to materials holding different properties. Their combination, by arbitrary stacking dangling-bond-free 2D nanosheets to form van der Waals heterostructures (vdWHs) 2 , enabled the further diversification of the properties 3,4 as a result of the charge redistribution at the interfaces of different 2DMs'crystals 5 . Artificial vdWHs can be produced by using diverse strategies, such as the pick-and-lift technique and wet-transfer from chemical vapor deposition (CVD) samples. However, all these methods suffer from the lack of control over the lattice alignment at the interface between two 2D nanosheets. Moreover, the transferred layers can exhibit microcracks, wrinkles, and contaminations by residues from the often-employed sacrificial layers. Epitaxial growth techniques are frequently exploited for the fabrication of 2D heterostructures via lateral overgrowth 6 . Additionally, lattice and thermal expansion coefficient mismatches limit the possibility for growth and integration of highefficiency electronic and photonic devices based on dissimilar 2DMs. Most importantly, neither of these conventional methods can meet the requirements for programming highly ordered electrostatic superlattices in vdWHs, to ultimately efficiently modulate the electronic properties of 2DMs. In this context, the use of the supramolecular networks represents a simple, yet remarkably effective solution to overcome the aforementioned problems. 2DMs provide atomically flat surfaces for molecular assembly governed by non-covalent interactions 7,8 . The physisorption of suitably designed molecules allows the formation of self-assembled monolayers (SAMs) under thermodynamic control which represents true superlattices on top of the crystalline 2DMs. The presence of electron-donating/withdrawing groups in the molecule employed to form the SAMs can lead to doping of 2DMs through the controlled local modifications of their surface potential [8][9][10] . Compared to molecules randomly adsorbed on the 2DMs, the use of highly ordered superlattices may maximize the electronics variation via collective or even cooperative effects. Pioneering works demonstrating the feasibility of 2DMs' surface decoration with molecular monolayers include organic/inorganic vdWHs of graphene, h-BN, and MoS 2 7 . While the band structures of the pristine layered materials are modified by the additional periodic potential introduced by SAM lattices, hitherto the examples are limited to monolayers formed via unspecific weak intermolecular interactions between alkyl chains that result in small domains (<10 5 nm 2 ) featuring random orientations [11][12][13] . Additionally, SAMs formed via lateral vdW interactions may undergo dynamic rearrangements upon humidity 14,15 or high temperature exposure 16 further leading to instabilities in 2DMbased (opto)electronic devices under operation. The use of 2D supramolecular structures held together by directional and specific non-covalent interactions appears as an ideal solution to attain higher stability and structural control over the micrometer scale. Intermolecular hydrogen bonds exhibit higher energy (25-40 KJ/mol) compared to intermolecular vdW forces (5 KJ/ mol), and they have been widely exploited to guide the selfassembly of suitably designed molecular modules into a variety of supramolecular architectures including 1D, 2D, and 3D arrangements 17 . This type of interaction is unique as it offers a high level of control over the molecular assembly process, along with high specificity, directionality, and reversibility. H-bonding interactions can benefit from a wide range of interaction energies, depending on the number and position of consecutive strong H-bonds. In this framework, melamine (M) represents an archetypal system for the formation of stable non-covalent architectures, in the form of bi-component networks with cyanuric acid (CA) 18 , perylene-3,4,9,10-tetracarboxylic diimide 19 , and other di-imide molecules 20,21 . Micrometer-size extended CA·M networks self-assembled on 2D substrates, including HOPG, Au(111), and black phosphorous [22][23][24][25] , have been shown to exhibit a low amount of structural defects and elevated thermal stability 23 . Here, we study the effect of the physisorbed CA·M bicomponent self-assembled network on the electronic properties of the underlying TMDCs. We found the occurrence of a strong charge-transfer process taking place on TMDCs as characterized by a positive cooperative effect emerging in densely packed CA·M bi-component tri-hapto hydrogen-bonded networks. Hybrid van der Waals heterostructures based on molybdenum disulfide (MoS 2 ) functionalized with CA·M supramolecular lattices exhibit an increase in charge carrier density, exceeding 10 13 cm −2 . Moreover, the same superlattice grown on tungsten diselenide (WSe 2 ) also displays similar significant improvement of the electronic transport through the 2D material. Further insights into the electronic effect of this functionalization approach are gained with density functional theory (DFT) calculations and Kelvin Probe Force Microscopy (KPFM). The efficient charge transfer from the physisorbed CA·M superlattice to the 2DMs determines an upward shift of the Fermi level (E F ) of the 2DM. The studies on MoS 2 as an earth-abundant and inexpensive material are essential for the implementation of clean energy technologies using hydrogen. The formation of MoS 2 based 2D heterostructures, e.g., graphene/MoS 2 26 and WS 2 /MoS 2 27 , induces a built-in electric field formed among the dissimilar layers, enhancing the MoS 2 catalytic performance. The highly crystalline nature of the 2D CA·M network and its significant n-doping effect on the 2DMs induces a strong enhancement of the hydrogen evolution reaction (HER) activity of MoS 2 , surpassing the performance of the standard polycrystalline platinum catalyst. The effect the CA·M superlattice shifts the HER overpotential of MoS 2 over 100 mV, yielding a small Tafel slope of 40 mV/dec. Results Positive cooperative effect on charge transfer. Cooperativity is a hallmark of molecular self-assembly, with the ensemble behaving differently as a whole from the sum of the isolated individual molecules. DFT calculations revealed that the CA and M coassembly on MoS 2 surface exhibits a positive cooperativity on the charge transfer process. Tightly packed 2D assemblies of M, CA, and CA·M onto the surface of monolayer MoS 2 have been modeled ( Supplementary Fig. 1). The different charge density images based on such superlattices (Fig. 1a-c) indicate the charge transfer process from the organic layer to the MoS 2 , and the calculated magnitude of charge transfer is summarized in Fig. 1d. M molecules are found to adopt a bent conformation in tightly packed monolayers on MoS 2 , which leads to a lower magnitude of charge transfer (2.47 × 10 13 cm −2 ) compared to nearly perfectly flat CA assemblies (2.70 × 10 13 cm −2 ). The bi-component CA·M assembly onto MoS 2 reveals a charge transfer magnitude as high as 3.62 × 10 13 cm −2 , hence outperforming the mono-component assemblies. To gain deeper insight into the role of the admolecules on the local electronics of the MoS 2 , the magnitude of the charge transfer is computed for mono-component M and CA as well as for the bi-component CA·M superlattice as a function of the tightness of the packing at the supramolecular level (Fig. 1e). We find a linear growth of the charge transfer with the decreasing area of the unit cell of the superlattice for monocomponent M and CA assemblies-based hybrids, highlighting a collective nature of the effect 28 The modified surface is then rinsed abundantly with water to remove the excess of physisorbed molecules not involved in the formation of the 2D CA·M network, followed by thermal annealing at 430 K in a vacuum for 12 h. Atomic force microscopy (AFM) imaging is employed to monitor the evolution of the surface morphology. It reveals that annealing under vacuum promotes molecular rearrangement on the MoS 2 surface yielding a drastic increase in the size of the CA·M domains up to tens of µm 2 forming a continuous crystalline film ( Supplementary Fig. 3b). Such 2D CA·M network exhibits an average thickness of 0.397 ± 0.103 nm. The structure of the dry self-assembled 2D CA·M network is further investigated with a sub-nm resolution by scanning NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-022-28116-y ARTICLE tunneling microscopy (STM) imaging under ambient conditions. In general, the on-surface molecular assembly is driven by the interplay between molecule-substrate and molecule-molecule interactions. In the present case, the 2D CA·M network formation is mainly governed by intermolecular interaction through the generation of three tri-hapto hydrogen bonds between one CA and three M molecules (Fig. 2a). Such self-assembly pattern further expands across the surface to form a large-area 2D supramolecular structure, where the CA·M network physisorbs onto the underlying MoS 2 support as stabilized by vdW interactions. Figure 2b reveals that the CA·M network displays a hexagonal symmetry similar to that of the underlying MoS 2 lattice. The unit cell parameters are measured as a = b = 0.96 ± 0.02 nm, and γ = 120 ± 2°. As the CA·M structure observed on the MoS 2 surface corresponds well to the periodic distance observed in bulk CA·M crystals (measured as 0.964 nm) 29 , we consider that the packing structure of the formed 2D CA·M monolayer is completely dominated by the intermolecular interactions. The detailed structural information gathered from high-resolution STM imaging (Fig. 2c) reveals the presence of hexagonal cavities in the CA·M network. DFT simulations of the optimized CA·M structure confirms that the CA·M network adsorbs flat on MoS 2 surface, with the CA and M molecules interacting through N-H···O and N-H···N hydrogen bonds ( Supplementary Fig. 1c). The simulated SAM structure perfectly matches the STM results. The same preparation procedure of this hybrid supramolecular adlayer was also applied by using the WSe 2 crystal as support, and a similar surface morphology and structure have been monitored by AFM and STM, respectively (Supplementary Fig. 4). Due to the strong multiple hydrogen bonds holding adjacent molecules together, the nature of the underlying surface has little effect on this bi-component assembly. The large-scale highly ordered CA·M monolayer can be easily obtained on the atomically flat surfaces, demonstrating that this should be a universal approach to prepare hybrid TMDC/CA·M vdWHs. Electrical and spectroscopic characteristics. The influence of the 2D CA·M network on the electronic characteristics of TMDCs has been investigated by means of electrical and spectroscopic characterizations. Monolayer MoS 2 back-gated FETs (Fig. 3a) are fabricated on n ++ -Si/SiO 2 (270 nm oxide thickness) with patterned top Au source and drain electrodes (channel length (L)/ width (W) amounting to 1.77 µm/1.79 µm, respectively). Transfer curves are recorded at source-drain bias V DS = 100 mV before and after the CA·M network deposition to assess its effect on the device electrical characteristics (Fig. 3b). Upon coating with CA·M monolayer, a significant shift of the MoS 2 FET threshold voltage (ΔV th ) to the negative V GS is observed ( Fig. 3b and Supplementary Fig. 5). The off-state current increases from 0 to 1.1 µA at V GS = −80 V. This can be attributed to a strong electrostatic n-doping effect 30 : the functionalized monolayer MoS 2 is degenerately doped showing reduced V GS dependence in the transfer curves. The n-doping effect can be quantified from the shift of the ΔV th , which linearly correlates to the charge carrier density change (Δn). Here, ΔV th is estimated as 125 V (Supplementary Fig. 5a) after CA·M formation, leading to a Δn of 1.1 × 10 13 cm −2 for the transfer curve shown in Fib. 2b. The measurements are reproducible on five different devices, reaching a maximum Δn of 2.7 × 10 13 cm −2 and average Δn of (1.9 ± 0.8) × 10 13 cm −2 . The calculated magnitude of the charge transfer matches well with the simulation result, 3.6 × 10 13 cm −2 in Fig. 1a. To our knowledge, previously reported changes in electron density caused by organic molecules physisorbed onto monolayer TMDCs falls usually in the range of 10 10 -10 12 cm −2 , even when thick molecular films are used [31][32][33][34] . Significantly, our results indicate that when the CA·M molecules are arranged into crystalline H-bonded self-assembled 2D supramolecular structures the doping effect is enhanced. Meanwhile, the field-effect electron mobility (μ) of the CA·M functionalized MoS 2 monolayer results ca. 38 cm 2 V −1 s −1 , being notably larger than the 14.3 cm 2 V −1 s −1 measured in pristine device. After CA·M coating, the increased electron density induces a decreased contact resistance and screening of Coulomb scatterings, yielding a μ increase. Further analysis of the device transfer curves indicate that hysteresis decreases from 27.5 V to 8.1 V upon CA·M coating ( Supplementary Fig. 5b, c), which is translated into a significant reduction of scattering from the interfacial traps (from 2.2 × 10 12 cm −2 eV −1 to 6.4 × 10 11 cm −2 eV −1 ) 35 . This provides the evidence for the screening of coulomb scattering from interfacial traps. Finally, the reversibility of our supramolecular functionalization strategy is confirmed by desorbing the molecules from the MoS 2 surface with hot water (>350 K), as evidenced by the FET transfer curves returning to the original state. Raman and PL spectroscopies are powerful techniques to assess and quantify the electronic properties of 2DMs and monitor fundamental processes like doping 36 . Here, we observed that the Raman signal from out-of-plane vibrations (A 1g ) of monolayer MoS 2 is sensitive to the presence of the CA·M superlattices. The E 1 2g mode at 386 cm −1 and A 1g mode at 405 cm −1 display a <20 cm −1 difference from the bare MoS 2 flake, confirming its monolayer nature 37 . CA·M coating results in a 1.3 ± 0.1 cm −1 redshift of the A 1g peak and nearly no change in the E 1 2g peak position (Fig. 3c). Doping in monolayer MoS 2 is reported to affect its A 1g vibration mode more significantly than E 1 2g , where A 1g redshift is a clear indication of the increase in the surface electron concentration of MoS 2 38 . The n-doping of MoS 2 upon CA·M coating is also confirmed by PL spectroscopy. For monolayer MoS 2 , the direct bandgap at K point enables strong PL emission 37,39 , as displayed as a black curve in Fig. 3d for the non-coated flake. Two major peaks originate from the spin-orbit splitting of transition metal Mo, viz. the A exciton peak at lower energies (≈1.84 eV) and the B exciton peak at higher energies (≈2.05 eV). Upon CA·M deposition, the MoS 2 A exciton peak redshifts of ca. 25 ± 2 meV and undergoes significantly quenching, while the B exciton peak remains unchanged. Our observations agree with the literature for chemical doping of monolayer MoS 2 by charge-transfer 37,38 . Similar results from the electrical and spectroscopical characterization of WSe 2 flakes confirm that the CA·M network acts as an n-type dopant on monolayer TMDCs (Supplementary Figs. 6 and 7). The reported assembly of H-bonded supramolecular networks as dopant of TMDC monolayers represents a simple, efficient, and reversible strategy to markedly improve device performance for various applications, such as in FETs and HER, as discussed hereafter. Doping and surface morphology relationships. Frequently, the doping of 2DMs is achieved upon physisorption of thick molecular layers 40,41 . Only a few previous works report the use of physisorbed SAMs to adjust the electrical properties of TMDCs, with the best Δn values amounting ca. 10 12 cm −2 12,13,32,42 . Here, the use of H-bonded 2D CA·M supramolecular networks yields an improvement of the electron density in monolayer MoS 2 , with carrier density being at least one order of magnitude superior. To shed light on the role played by the CA·M monolayer on such doping, we studied the morphology of pristine and functionalized device's channel by AFM (Supplementary Fig. 8). The intermolecular multiple H-bonds determine a high stability of the CA·M 2D structure, which grows forming highly ordered molecular film with micrometer-scale crystalline domains, fully covering the TMDC flake's surface. The surface roughness displays a very minor increase upon the physisorption of the CA·M film on both TMDCs, with values of 0.81 nm onto MoS 2 monolayer and 0.68 nm onto WSe 2 bilayer as quantified on a 1 µm 2 area. The defect-free, long-range crystalline structure of CA·M turns out to be crucial to promote such elevated doping. CA·M films produced at lower annealing temperatures (373 K instead of 430 K) onto WSe 2 FETs assemble into several smaller crystalline domains (of ca. 0.25 µm 2 ), resulting in a weaker doping effect (Supplementary Fig. 9). Additionally, we analyze the contributions of the individual components of the supramolecular network, i.e., M and CA molecules on the morphological and electrical characteristics of TMDCs (Supplementary Figs. [10][11][12][13]. Individual M molecules are observed to self-assemble into multiple small domains on MoS 2 (with an average size of 2.5 × 10 5 nm 2 ), which are easily disassembled during thermal annealing ( Supplementary Fig. 11). On the other hand, individual CA molecules form aggregates on the MoS 2 surface at room temperature ( Supplementary Fig. 12). From AFM phase images, while the molecular film of individual M covers almost half of the device channel's surface, CA molecules display a greater propensity to undergo aggregation, leaving most of the flake surface uncoated. Such substantial dissimilarity in morphology between mono-component and bi-component films determines a different influence on the electrical characteristics of MoS 2 based FETs. For M-only coated devices the negative V th shift is more pronounced than using CA molecules, viz. 28.1 V and 3.4 V, respectively ( Supplementary Fig. 13). This leads to Δn of 2.2 × 10 12 cm −2 for M-only and 3.7 × 10 11 cm −2 for CA-only coatings on the MoS 2 FETs. For measurements taken on five similar devices, the average Δn amounts to (1.9 ± 0.3) × 10 12 cm −2 and (4.7 ± 1.5) × 10 11 cm 2 for M-only and CA-only, respectively, which are both significantly smaller than the values measured on H-bonded CA·M networks with a Δn = (1.9 ± 0.8) × 10 13 cm −2 . The tight packing of CA·M monolayer on TMDC flakes may also act as an encapsulation layer to protect the device from the environment and subsequent degradation of performance. Density of states and Work function modulation. Further insights into the electronic properties of the MoS 2 /CA·M hybrid system can be obtained by elucidating the DOS and WF. The doping of a semiconductor shifts its E F position, ultimately influencing the material's WF 43,44 . Alongside, E F of a 2D semiconductor is particularly sensitive to modification with organic molecules 45,46 . E F is known to shift towards the conduction band minimum (CBM) in the case of n-type doping, leading to a lower WF, while p-type doping causes the opposite effect by shifting E F towards the valence band maximum (VBM). The evolution of the WF of a monolayer MoS 2 surface is measured by Kelvin probe force microscope (KPFM), which allows to record a contact potential map. As shown in Fig. 4a, d, freshly evaporated Au electrodes are used to ground the single MoS 2 flakes deposited on a 270 nm SiO 2 substrate. Darker regions in the KPFM images (Fig. 4b, e) correspond to larger WF values, indicating that the WF of both pristine and CA·M-doped MoS 2 surfaces is smaller compared to the Au electrodes, as shown by the brighter contrast of the flakes. The contact potential difference related to Au increases from 50 mV to 170 mV with CA·M coating. The presence of the CA·M network decreases the WF by 120 meV compared to the pristine MoS 2 surface, evidencing a n-doping effect, as determined by averaging the SP values on five monolayer MoS 2 flakes and six CA·M coated monolayer MoS 2 samples (Supplementary Table 1). Note here that SP differences of nano-objects determined via amplitude modulation KPFM are usually underestimated due to the finite size of the probe, which limits the spatial and potential resolution of the measurement 47 . The calculated WF values ( Supplementary Fig. 14a) revealed a negative shift in WF by 220 meV, in good agreement with the KPFM results. These results indicate that due to the interaction between the CA·M and MoS 2 surface, the removal of an electron from the whole hybrid system is much easier than from the CA·M SAM or MoS 2 surface separately. The DOS simulations are also in good agreement with the literature 44 and corroborate the experimental results, proving that the CA·M network determines n-type doping on monolayer MoS 2 surface. The band structure maps are shown in Fig. 4c, f: the energy gap from E F to CBM decrease from 1.479 to 1.249 eV after introducing the CA·M SAM. The upward shift of E F leads to an increased probability of exciting electrons to CBM. The n-type doping on MoS 2 mostly originates from the charge transfer effect of CA.M molecular network, which can be read from the partial density of states (PDOS) plots-derived energy level map ( Supplementary Fig. 15). The MoS 2 monolayer flakes coated with the individual molecular components (either M or CA) were also measured via KPFM, as depicted in Supplementary Fig. 17. Coating with M causes a reduction of the WF of pristine MoS 2 of ΔWF = −80 meV, thus being of smaller magnitude compared to the CA·M system. The WF does not change significantly in CA-coated flakes because of the propensity of the molecules to undergo strong aggregation. When coated with a CA·M superlattice the bilayer WSe 2 displays a negative WF shift of 95 meV ( Supplementary Fig. 18), which in good agreement with the simulated monolayer WSe 2 , exhibiting a ΔWF = −170 meV (Supplementary Fig. 14b). The experimental WF results from KPFM are summarized in Supplementary Table 1. High-efficiency HER catalytic performance. Electrochemical water splitting is considered one of the most promising approaches towards the production of hydrogen fuel. Platinum (Pt) is so far the best performing and most employed electrocatalysts for the hydrogen evolution reaction (HER) 48 , yet its high cost and scarcity greatly limit large-scale applications. Among various possible alternatives to Pt, MoS 2material abundant in nature -exhibits Gibbs free energy adequate for atomic hydrogen adsorption 49 . Although the MoS 2 catalytic activity has been significantly improved over the last decade 50,51 , it is still far from the HER performance of Pt. The catalytic activity of MoS 2 is known to be affected by its interaction with supporting substrates via charge transfer 52,53 . CA·M co-crystal nanosheets have been reported as supports for CoP nanoparticles catalyst, giving embedded composite with high hydrogen evolution activity and stability 54 . Since the CA·M network has been proved as a strong n-dopant to MoS 2 , it is also expected to act as a suitable buffer layer to boost the catalytic performance of MoS 2 . In HER, the adsorbed protons sequester electrons from the MoS 2 surface thereby forming hydrogen molecules through a process of desorption from the surface. Therefore, the electron density on the MoS 2 surface can indirectly affect the rate of hydrogen evolution to a certain extent. The electron transfer in the CA·M/MoS 2 heterostructure occurs from CA·M to MoS 2 , hence increasing the charge density on MoS 2 surface and further enhancing the HER activity. Multilayer CA·M film is prepared by drop-casting 200 µL of CA and M mixture solution at 1 × 10 −6 M concentration on a 1 cm 2 270-nm-thick SiO 2 wafer, and then annealed overnight under vacuum at 450 K. The obtained CA·M multilayer film exhibits roughness of 0.29 nm, allowing the easy deposition of mechanically exfoliated monolayer MoS 2 flake on its surface ( Supplementary Fig. 19). Raman and PL characterizations confirm that upon inverting the position of the supramolecular network with respect to MoS 2 (i.e., MoS 2 on CA·M) a strong n-doping effect is still induced on the upper MoS 2 monolayer. The typical three-electrode electrochemical system is employed to evaluate the HER performance of CA·M/MoS 2 . A micron-sized window is opened on a photoresist passivation layer on the device surface to expose only the MoS 2 flake to the electrolyte (0.5 M H 2 SO 4 ). Here, monolayer MoS 2 flake is used as working electrode (WE), while Ag/AgCl and pencil graphite are used as reference (RE) and counter (CE) electrodes, respectively (see Fig. 5a and Supplementary Fig. 20 for details). The catalytic performance of neat monolayer MoS 2 flake shows a clear improvement upon addition of the underlying CA·M thin film, as exhibited in Fig. 5c. The HER overpotential (η) is defined as the potential at the standard current density (j) −10 mA cm −2 . Here, the CA·M/ MoS 2 hybrid device reaches η = 208 mV,~130 mV smaller than that for the pristine MoS 2 surface (η = 340 mV), indicating a great enhancement of energy conversion efficiency. The participation of CA·M film improves the electrical coupling between the substrate and the MoS 2 catalyst, therefore the injection of electrons from the electrode and their transport to the catalyst active site is facilitated. Moreover, the CA·M strong n-doping effect induces an excessive negative charge density on the MoS 2 surface that boosts the HER performance. Figure 5d depicts the Tafel slope, an important parameter that quantifies the amount of additional applied potential is required to observe a logarithmic increase in current density during HER. The CA·M/MoS 2 hybrid device shows an Tafel slope of 40 mV/dec, which is significantly superior to the values reported for most 2H MoS 2 related compounds (typically 41-120 mV/dec) 55 , and even comparable with that for metallic 1 T MoS 2 (43 mV/dec) 56 . The integrity of the MoS 2 flakes is confirmed by Raman spectroscopy carried out after the HER measurements, which endorses the material as a very stable high-performing catalyst. The conductivity of the MoS 2 channel undergoes a sharp increase when a positive V GS is applied to the FETs, which is considered being a powerful strategy for enhancing the HER catalytic performance 57,58 . The FET characteristics of CA·M/ MoS 2 validates that the bottom multilayer CA·M film can induce significant electron density increases on MoS 2 upon back-gate V GS modulation ( Supplementary Fig. 21). Moreover, when MoS 2 surface is immersed in a sulfuric acid electrolyte, the applied positive gate voltage has also been confirmed to induce the excess electron density on the Mo atoms near the S vacancy sites, leading to enhanced Mo-H bond strengths (stabilizing H proton adsorption) 59 . The HER catalytic property of our field-tuned CA·M/MoS 2 devices is largely boosted with V GS increasing ( Supplementary Fig. 23c). In particular, the CA·M/MoS 2 channel achieves a better performance than standard polycrystalline Pt catalyst when the CA·M/MoS 2 vdWHs is gated at V GS = +4.3 V, exhibiting a η = 14 mV (20 mV for Pt) and a small Tafel slope of 21 mV/dec (30 mV/dec for Pt). Interestingly, the field effect on the bare monolayer MoS 2 surface is much weaker than on the CA·M/MoS 2 hybrid surface. When applying V GS from 0 V to +4 V, the over potential decreases by 62 mV for monolayer MoS 2 and by impressive 180 mV for the CA·M/MoS 2 hybrid structure (Fig. 5c). The strong n-doping effect from the underlying CA·M film enables directional electron flow to MoS 2 active sites, which determines the excellent catalytic performance of this hybrid structure to produce molecular hydrogen. Such a strategy of utilizing the electric field to improve the catalytic activity of HER could also be used in the other electrochemical processes. However, the use of more complicated four electrodes system in cyclic voltammetry experiments may hinder to some extent this gate-enhanced HER strategy for large-scale technological applications. Discussion vdWHs based on 2DMs is attracting ever-growing attention due to their unique electronic features that can be engineered on demand. In this context, the use of the small organic building blocks capable of forming supramolecular lattices through noncovalent interactions represents a simple, yet effective solution towards the modulation of 2DM electronic properties. Here, we show that the bi-component, hydrogen-bonded CA·M 2D network can be used for boosting the electron density of TMDCs through a positive cooperative effect in the charge transfer process as a function of the packing density of the molecular adlayer. We have found that the physorption of a tightly packed 2D supramolecular network of CA·M onto the monolayer MoS 2 channel determines an electron transfer up to 2.72 × 10 13 cm −2 , thereby reaching state-of-the-art performance compared to the reported molecule-TMDCs hybrid devices. Optical and electrical analyses corroborated with DFT calculations reveal that the CA·M 2D network induces a charge transfer effect to MoS 2 , regardless if placed on top or underneath the 2DMs. Owing to the excellent charge transfer effect, CA·M 2D network is used as a substrate to enhance the HER catalyst performance of monolayer MoS 2 . Remarkably, the field-effect enhanced HER performance of the CA·M/MoS 2 hybrid device outperforms the catalytic efficiency of Pt. 2DMs decorated with supramolecular 2D networks may expand the library of heterostructures with on-demand electronic characteristics. Moreover, such an approach can be further employed in multilayer vdWHs displaying electrical, optical, and magnetic functionalities. Compared with classical vdW 2DMs heterostructures, the structural programmability of 2D supramolecular networks enables optimal matching of the crystal structures to form ad hoc superlattices. Importantly, supramolecular chemistry can offer a choice of 2D networks and patterns, with tunable electronic properties, periodic potentials, and atom packing density. The atomic precision achieved through controlled molecular self-assembly can therefore be instrumental for the optimization of charge carriers in 2DMs, and be used to impart novel properties, enabling the realization of multifunctional, high-performance optoelectronic devices. Methods STM and AFM. STM measurements were carried out by using a Veeco scanning tunneling microscope (multimode Nanoscope III, Veeco). Bulk MoS 2 and WSe 2 crystals (HQ graphene) were used as substrates before and after CA·M coating. The employed STM tips were mechanically cut from a Pt/Ir wire (80/10, diameter 0.25 mm). The images were obtained at room temperature in air, and the drift of the piezo was corrected using the underlying chalcogen atomic lattice as a reference. AFM imaging was performed using a Bruker Dimension Icon set-up operating in air, in tapping mode, by using tip model TESPA-V2, tip stiffness K is 42 N m −1 . Raman and photoluminescence spectra. Raman spectra were carried out in air by Renishaw inVia spectrometer equipped with 532 nm laser, and the wavenumber resolution was 1 meV. The Si Raman peak at 520.5 cm −1 was used for normalization. The PL spectra were acquired with a Renishaw InVia spectrometer equipped with a 532 nm laser. TMDC flakes were mechanically exfoliated from crystals onto SiO 2 /Si substrate. The excitation power was kept below 1 mW to avoid the local heating and damaging effect. FET fabrication and characterization. MoS 2 and WSe 2 flakes were mechanically exfoliated down to monolayer by the Scotch-tape-based method and further transferred to CA·M/SiO 2 and bare SiO 2 substrates (270 nm SiO 2 thickness onto ndoped Si from Fraunhofer IPMS, Germany). Top-gated FET devices were fabricated by maskless optical lithography, with Au source and drain electrodes thermally evaporated onto the patterned substrate. Warm acetone was used for the liftoff process, and the as-fabricated devices were annealed at 430 K in vacuum to remove atmospheric adsorbates. All the FET devices were measured in an N 2 -filled glovebox with a probe station connected to a Keithley 2636 A source-meter unit. The carrier mobility µ was determined from the following Eq. (1): L and W are the channel length and width, and C i is the insulator capacitance per unit of area. The dopant-induced charge carrier density increase (Δn) was calculated from Eq. (2): Where ΔV th is the difference in threshold voltage (V th ) before and after doping effect, and e is the elementary charge (1.6 × 10 −19 C). KPFM. Topography and surface potential images were simultaneously collected with Pt/Ir coated silicon probes (Bruker SCM-PIT-V2, resonant frequency ≈75 kHz, k ≈ 3 N·m −1 ) at ambient conditions using a Bruker Icon AFM employed in amplitude modulation KPFM (AM-KPFM) mode. For work function (WF) referencing and electrical grounding purposes, a polycrystalline Au electrode was thermally evaporated with a shadow mask on the flakes, before and after the molecule adsorption. The reported surface potential values are referred to the average SP of the non-treated Au electrode (0.16 ± 0.10 eV) obtained from >30 samples. The calibration of the Au WF (4.85 ± 0.09 eV) was determined on the polycrystalline Au electrodes via macroscopic Kelvin Probe (KP) at ambient conditions (Ambient Kelvin Probe Package from KP Technology Ltd, 2-mm-diameter gold tip amplifier) 60 . The calibration of the KP probe was performed against a freshly cleaved HOPG surface (4.475 eV). The SPM image processing has been done on WSxM software. Device fabrication for HER measurements. Devices for HER measurements were produced by spin coating a layer of photoresist (AZ1505) onto monolayer MoS 2 transferred to n ++ -Si/SiO 2 substrate. A window was opened on the photoresist by mask-less optical lithography to expose the desired flake area to the solution during measurements. The HER catalytic performance of monolayer MoS 2 was measured in a three-electrode configuration in 0.5 M H 2 SO 4 electrolyte (purged with Ar gas), with an Ag/AgCl reference electrode and pencil graphite counter electrode (diameter 3 mm). The MoS 2 flake acts as the working electrode during linear sweep voltammetry (LSV) from 0.1 to −0.6 V, with a scan rate of 3 mV s −1 . All LSV curves were collected after 5 times scan in the same potential range. HER measurements were also performed while biasing the n ++ -Si gate electrode, as depicted in Fig. 4a. The electrochemical current density is calculated by normalizing the current to the area of the exposed window on the MoS 2 . The data has been referenced to E RHE = E Ag/AgCl + 0.210 V. DFT calculations. First-principles calculations within the density functional theory (DFT) framework were performed to investigate different types of self-assembled monolayers (CA and M) adsorbed on the MoS 2 surface structures. By means of DFT calculations the adsorption of CA·M SAM on the different surface structures was investigated. Here an important role was played by the small distance (~3.3 A) between the SAM and the surface, which can influence the electronic properties of the whole system. The WFs were simulated by DFT. In order to identify the most stable configurations, we performed first-principles total-energy calculations using the gradient corrected (PBE) density functional theory as implemented in the Vienna ab initio simulation package 61,62 . The geometry of the MoS 2 /CA·M hybrid structure was optimized by relaxing the atomic positions of all the atoms. The equilibrium geometry was assumed to be reached when the forces on the relaxed atoms of the system were less than 0.025 eV/Å. A (2 × 2 × 1) Monkhorst-Pack k-point mesh has been used to span the surface Brillouin zone. The system was modeled by using a periodic supercell of 9.81 × 9.81 × 18 Å 3 . The bond length parameters as well as the unit cell of the structure were chosen to fit the experimental observations. The electronic wave functions were expanded into plane waves up to an energy cutoff of 400 eV, and a projected-augmented-wave scheme 63 was used in order to describe the interactions between the valence electrons and the nuclei(ions). Data availability	2023-02-09T15:16:57.205Z	2022-01-26T00:00:00.000	{ "year": 2022, "sha1": "6d1e513de7bd435fb01e53000728c8ada8e4d29c", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41467-022-28116-y.pdf", "oa_status": "GOLD", "pdf_src": "SpringerNature", "pdf_hash": "6d1e513de7bd435fb01e53000728c8ada8e4d29c", "s2fieldsofstudy": [ "Chemistry" ], "extfieldsofstudy": [] }
4639960	pes2o/s2orc	v3-fos-license	Comparison of Immunohistochemistry and Direct Sanger Sequencing for Detection of the BRAFV600E Mutation in Thyroid Neoplasm Background The BRAFV600E mutation is the most common genetic alteration identified in papillary thyroid carcinoma (PTC). Because of its costs effectiveness and sensitivity, direct Sanger sequencing has several limitations. The aim of this study was to evaluate the efficiency of immunohistochemistry (IHC) as an alternative method to detect the BRAFV600E mutation in preoperative and postoperative tissue samples. Methods We evaluated 71 patients who underwent thyroid surgery with the result of direct sequencing of the BRAFV600E mutation. IHC staining of the BRAFV600E mutation was performed in 49 preoperative and 23 postoperative thyroid specimens. Results Sixty-two patients (87.3%) had PTC, and of these, BRAFV600E was confirmed by direct sequencing in 57 patients (91.9%). In 23 postoperative tissue samples, the BRAFV600E mutation was detected in 16 samples (70%) by direct sequencing and 18 samples (78%) by IHC. In 24 fine needle aspiration (FNA) samples, BRAFV600E was detected in 18 samples (75%) by direct sequencing and 16 samples (67%) by IHC. In 25 core needle biopsy (CNB) samples, the BRAFV600E mutation was detected in 15 samples (60%) by direct sequencing and 16 samples (64%) by IHC. The sensitivity and specificity of IHC for detecting the BRAFV600E mutation were 77.8% and 66.7% in FNA samples and 99.3% and 80.0% in CNB samples. Conclusion IHC could be an alternative method to direct Sanger sequencing for BRAFV600E mutation detection both in postoperative and preoperative samples. However, application of IHC to detect the BRAFV600E mutation in FNA samples is of limited value compared with direct sequencing. INTRODUCTION The BRAF V600E mutation is the most common mutation detected in papillary thyroid carcinoma (PTC).In Korea, PTCs harboring BRAF V600E were reported in up to 80% of the population [1][2][3].The BRAF oncogene encodes the human gene for B-type Raf kinase.BRAF V600E is a T1799A point mutation in exon 15, resulting in the substitution of valine to glutamate at residue 600 (V600E).It activates the RAS/RAF/MAPK signaling pathway by disrupting hydrophobic interactions between residues in both the activation loop and the ATP-binding site [4].BRAF V600E blocks apoptosis and regulates proliferation and invasiveness [5].In addition, BRAF V600E is associated with aggressive clinical behavior and a higher risk of recurrence of PTC [6][7][8] and may help in detecting malignancy in some thyroid nodules [9,10].The BRAF V600E mutation is one of the most stressed molecular markers for its clinical role in thyroid cancer. The gold standard method for molecular diagnosis of the BRAF V600E mutation is direct Sanger sequencing, which is highly reliable and specific [11].However, the sensitivity of Sanger sequencing was reported to identify only 7% to 20% of mutated alleles in a background of wild-type alleles [11,12].The tumor content of malignant cells is reported to influence the sensitivity of Sanger sequencing [13].Over the last decade, several alternatives to Sanger sequencing for detecting the BRAF V600E mutation have been reported with different sensitivities and specificities [11,[14][15][16]. Recently, immunohistochemistry (IHC) using a BRAF V600Especific antibody (VE1) has been used to detect the BRAF V600E mutation [17].IHC is a widely available and low-cost technique.Previous studies suggest that IHC is highly sensitive for detecting the BRAF V600E mutation [18][19][20]. In the present study, the efficiency of BRAF V600E IHC as an alternative method to direct Sanger sequencing was evaluated in preoperative and postoperative thyroid tissue samples. Tissue samples and study design Seventy-one patients who underwent thyroid surgery between January 2011 and January 2013 in the Asan Medical Center, Seoul, Korea, were included in this retrospective study.Results on the BRAF mutation status using direct Sanger sequencing were available.Diagnostic preoperative tissue samples were obtained by either fine needle aspiration (FNA) or core needle biopsy (CNB).Postoperative formalin-fixed, paraffin-embedded (FFPE) samples were obtained after thyroid surgery.IHC staining for the BRAF V600E mutation was performed in preoperative and postoperative thyroid specimens.The results of BRAF V600E IHC were compared with those of direct Sanger sequencing.The study protocol was approved by the Institutional Review Board of the Asan Medical Center (AMC 2016-0821).Informed consent was obtained from all patients for BRAF V600E mutation analysis. Cytological and histopathological diagnosis All thyroid specimens were reviewed by an experienced endocrine pathologist (D.E.S.).FNA cytology results were categorized into six categories using the Bethesda system [21].Preoperative CNB specimens were evaluated based on the criteria proposed by the Korean endocrine pathology thyroid CNB study group [22].Surgical specimens were reviewed and diagnosed based on the World Health Organization classification criteria [23]. Ultrasonography-guided fine needle aspiration and core needle biopsy procedures Neck ultrasound (US) examinations were performed using either an iU22 unit (Philips Healthcare, Bothell, WA, USA) or an EUB-7500 unit (Hitachi Medical Systems, Tokyo, Japan) equipped with a linear high-frequency probe (5 to 14 MHz), as previously reported [24].The scanning protocol included both transverse and longitudinal real-time imaging.All US examinations and FNA/CNB procedures were performed by experienced radiologists.FNA procedures were performed under US guidance using a 23-gauge needle connected to a 10 mL syringe [25].US-guided CNB procedures were performed using an 18-gauge, 1.1 or 1.6 cm excursion, double-action, spring-activated needle (TSK Ace-cut, Create Medic, Yokohama, Japan) [26,27]. Immunohistochemistry for the BRAF V600E mutation IHC for the BRAF V600E mutation was performed from 49 preoperative FNA cell blocks or CNB specimens and all postoperative thyroidectomy specimens using mouse anti-BRAF V600E (clone VE1, 1:4 dilution, Ventana Medical Systems, Tucson, AZ, USA) and the OptiView DAB IHC Detection Kit (Roche, Tucson, AZ, USA) on a Ventana BenchMark XT autoimmunostainer (Ventana Medical Systems) [28].Whole tissue sections (4 µm) were transferred to poly-L-lysine-coated adhesive slides and dried at room temperature for 10 minutes.Antigen retrieval was performed using the heat-induced epitope retrieval method, Copyright © 2018 Korean Endocrine Society followed by 20 minutes at 65°C in an incubator.After incubation in CC1 solution (Roche) for 32 minutes, primary antibody was added and incubated for 16 minutes.A haptenated secondary antibody was added and incubated for a further 10 minutes.Alpha-hapten-horseradish peroxidase was added to the sections.After incubation for 10 minutes, diaminobenzidine was added and incubated for 10 minutes.Negative controls were performed by omitting the primary antibody.Positive controls were melanoma tissue. Slides were evaluated by an experienced pathologist (D.E.S.) who was blinded to the BRAF mutational status.Diffuse homogeneous cytoplasmic staining in all tumor cells was considered as positive.Non-specific staining of colloids and equivocally weak or focal cytoplasmic staining was considered as negative. Direct Sanger sequencing for BRAF Sanger sequencing was performed as previously described [27].Briefly, exon 15 of the BRAF gene was amplified by polymerase chain reaction (PCR) using the primers 5´-TGCTT-GCTCTGATAGGAAAATG-3´ (forward) and 5´-CTGATGGG ACCCACTCC AT-3´ (reverse).The amplification protocol comprised initial denaturation at 94°C for 2 minutes; 40 cycles of denaturation at 95°C for 10 seconds, annealing at 60°C for 30 seconds, and extension at 68°C for 40 seconds; followed by a final extension at 68°C for 7 minutes using KOD FX Taq DNA polymerase (Toyobo, Osaka, Japan).PCR products were electrophoresed on 2.5% (wt/vol) agarose gels and subsequently sequenced using the above forward primer and Big Dye Terminator (ABI Systems, Applied Biosystems, Foster City, CA, USA).DNA sequences and the BRAF mutation were determined using an ABI-PRISM 3100 automatic sequencer (Applied Biosystems). Statistical analysis Statistical analyses were performed using R version 3.0 and the R libraries (R Foundation for Statistical Computing, Vienna, Austria, http://www.R-project.org).Categorical variables are presented as numbers with percentages.McNemar's chi-square test was used to evaluate the performance of two tests.All P values were two-sided and P values <0.05 were considered statistically significant. Baseline characteristics of study subjects and tissue samples Sixty-two of 71 patients (87.3%) had PTC including one patient with a follicular variant of PTC, and of these, the BRAF V600E mutation was detected in 57 patients (91.9%) by direct sequencing.Four patients (5.6%) had follicular thyroid carcinoma, and an additional four patients (5.6%) had nodular hyperplasia.Wild-type BRAF was confirmed in 14 tumors (19.7%). Comparison of direct Sanger sequencing and IHC results for the BRAF V600E mutation using postoperative samples Table 1 shows the results of direct Sanger sequencing and IHC staining for the BRAF V600E mutation in 23 postoperative FFPE samples.The BRAF V600E mutation was detected in 16 samples (70%) using direct sequencing.IHC for BRAF V600E was positive in 18 FFPE samples (78%).Of these samples, two showed discrepant results between the two methods.IHC results were positive, whereas the results of direct sequencing showed wild-type BRAF (Fig. 1A, B).The sensitivity and specificity of IHC for detecting the BRAF V600E mutation in postoperative samples were 100% and 71.4%.There was no significant difference in detecting the BRAF V600E mutation between two tests in postoperative samples (P=0.480). Comparison of direct Sanger sequencing and IHC results for the BRAF V600E mutation using preoperative samples Table 2 shows the results of direct Sanger sequencing and IHC staining for the BRAF V600E mutation in 49 preoperative samples including 24 FNA samples and 25 CNB samples.The BRAF V600E mutation was detected in 33 samples (67%) using direct sequencing and 32 samples (65%) using IHC.In preoperative samples, inconsistent results between the two methods were detected in nine patients (18.4%).Of these, four were positive by IHC and negative by direct sequencing.Five patients were negative with IHC and positive with direct sequencing methods.The sensitivity and specificity of IHC for detecting the BRAF V600E mutation in preoperative samples were 84.9% and 75.0%.No difference was found between two tests in detecting Copyright © 2018 Korean Endocrine Society the BRAF V600E mutation in preoperative samples (P=0.999). In 24 FNA specimens, the BRAF V600E mutation was detected in 18 samples (75%) by direct sequencing and 16 samples (67%) by IHC.The results of the two methods were discordant in six of 24 samples (25%).Two samples were positive for the BRAF V600E mutation by IHC, but direct sequencing results showed wild-type BRAF.The BRAF V600E mutation was detected in four samples using direct sequencing, but these were negative using IHC.BRAF V600E mutations from two samples using direct sequencing are shown in Fig. 1C, D. However, results of IHC were negative in the two corresponding FNA cell blocks as shown in Fig. 1E, F. The sensitivity and specificity of IHC for detecting the BRAF V600E mutation in FNA samples were 77.8% and 66.7%.There was no statistical significant difference in detecting the BRAF V600E mutation between two tests in FNA samples (P=0.683). In 25 CNB specimens, the BRAF V600E mutation was detected in 15 samples (60%) by direct sequencing and 16 samples (64%) by IHC.The results of direct sequencing and IHC differed in three of 25 samples (12%).The BRAF V600E mutation was detected in one sample by direct sequencing, but was negative in IHC.Two samples were positive for the BRAF V600E mutation using IHC, but were not detected by direct sequencing (Fig. 1G, H).The sensitivity and specificity of IHC for detecting the BRAF V600E mutation in CNB samples were 99.3% and 80.0%. There was no significant difference in detecting the BRAF V600E mutation between two tests in CNB samples (P=0.999). DISCUSSION The presence of the BRAF V600E mutation is considered an intermediate risk factor in the recently revised American Thyroid Association guideline for thyroid nodules and differentiated thyroid cancer [29].This indicates that the presence of BRAF V600E is generally considered as a risk factor for poor prognosis or aggressive behavior of thyroid cancer.BRAF V600E mutation analysis is difficult to perform in every thyroid cancer patient, because of the issue of cost-effectiveness. The efficiency of IHC, a low-cost method, for detecting the BRAF V600E mutation was evaluated as an alternative technique to direct Sanger sequencing in different preoperative and postoperative thyroid tissue samples.The sensitivity of IHC staining for detecting the BRAF V600E mutation was found to be superior to that of direct Sanger sequencing in postoperative tissue samples.In preoperative tissue samples, IHC staining performed better in CNB samples. The BRAF V600E mutation was found in 92% of PTC patients in the present study.This result is consistent with findings of previous studies demonstrating a relatively high prevalence of the BRAF V600E mutation in Korean PTC patients [1,30].This implies that detection of the BRAF V600E mutation could play a role in the diagnosis or management of PTC patients, especially in Korea. According to our data, the concordance between direct Sanger sequencing and IHC for detecting BRAF V600E was superior in postoperative tissue samples compared with preoperative tissue samples.Discordance between the two methods in postoperative tissue samples was found in only two patients, who were wild-type by direct Sanger sequencing and positive using IHC.In preoperative tissue samples, discordant results between the two methods were found in nine samples, of which four were wild-type by direct Sanger sequencing and positive using IHC, and the remaining five were vice versa.These results imply that IHC is more suitable for postoperative tissue samples than preoperative tissue samples. Preoperative tissue samples were obtained by FNA and CNB.Data showed that CNB samples were more suitable for IHC staining than FNA samples, even though there were no statistical differences between Sanger sequencing and IHC staining methods on both tissue samples.According to previous studies, IHC on FNA samples shows relatively less specificity and sensitivity than on FFPE samples [18,31,32].However, Crescenzi www.e-enm.org67 et al. [33] reported that IHC performed on thyroid CNB samples perfectly matched genetic analysis of BRAF status.These data support the use of IHC on CNB samples to detect BRAF V600E at preoperative evaluation of thyroid nodules.Five tissue samples were positive by direct Sanger sequencing and negative by IHC, with four FNA samples and one CNB sample.All of these samples were collected preoperatively.These findings may be due to loss of antigen expression of the mutation.For example, tissue ischemic areas surrounding necrotic areas or damaged tissue have reduced BRAF V600E protein expression, which may be the cause of the false-negative IHC results, particularly in small biopsies such as FNA [18,34]. As shown in Supplemental Table S1, the FNA sample from subject number 1 showed negative results by direct Sanger sequencing and positive results by IHC.In this patient, subsequent reverse transcriptase-PCR revealed the BRAF V600E mutation.This implies that methods for detecting the BRAF V600E mutation may affect the results of mutation analysis.Combination of IHC and molecular approaches has been suggested for better detection of the BRAF V600E mutation [35].However, the cost-effectiveness of using these approaches requires consideration. In conclusion, IHC could be an alternative approach to direct Sanger sequencing for detecting the BRAF V600E mutation both in postoperative and preoperative tissue samples.However, application of IHC in FNA samples was of limited value compared with direct sequencing.Direct Sanger sequencing is the gold standard method for detecting the BRAF V600E mutation and is highly specific.Disadvantages include its relative low sensitivity compared with other tests [11,35] and high costs.IHC on appropriate tissue samples is an alternative and superior approach in detecting the BRAF V600E mutation because of its improved performance and low cost. Fig. 1 . Fig. 1.Detection of the BRAF V600E mutation in different thyroid tissue specimens with inconsistent results between direct Sanger sequencing and immunohistochemistry (IHC) (A-F).Two postoperative thyroidectomy specimens were positive by immunohistochemistry for the BRAF V600E mutation (A, B, ×400), but negative by direct Sanger sequencing.Two preoperative fine needle aspiration cell block specimens showed discrepant results with positive direct Sanger sequencing (C, D) results and negative IHC results (E, F, ×400).Two preoperative core needle biopsy specimens were positive by IHC for the BRAF V600E mutation (G, H, ×400), but negative by direct Sanger sequencing. Table 1 . Comparison of the Results of Direct Sanger Sequencing and IHC for BRAF V600E in Postoperative Samples Table 2 . Comparison of the Results of Direct Sanger Sequencing and IHC for BRAF V600E in Preoperative Samples	2018-04-03T05:17:45.171Z	2018-01-30T00:00:00.000	{ "year": 2018, "sha1": "35fee07d42641f5c995ec19d5ee41cd44f83461f", "oa_license": "CCBYNC", "oa_url": "https://doi.org/10.3803/enm.2018.33.1.62", "oa_status": "GOLD", "pdf_src": "ScienceParseMerged", "pdf_hash": "35fee07d42641f5c995ec19d5ee41cd44f83461f", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
261608630	pes2o/s2orc	v3-fos-license	Assessing the Adequacy of Superficial Temporal Artery Blood Flow in Korean Patients Undergoing STA-MCA Anastomosis Objective Superficial temporal artery (STA)-middle cerebral artery (MCA) anastomosis is conducted for flow augmentation. In this study, we measured the STA cut flow of a Korean population and evaluated the relationship between STA cut flow and long-term patency of the bypass. Methods A retrospective study was conducted. Intraoperative measurement of STA flow was conducted using a microvascular flow meter on patients who underwent STA-MCA. After cutting the distal end, the STA flow rate was measured with no resistance and recorded. After finishing anastomosis, STA flow was measured and recorded. The cut flow index was calculated by dividing post anastomosis flow by cut flow in intracranial atherosclerotic stenosis patients. Results The median STA cut flow was 35.0 mL/min and the post anastomosis flow was 24.0 mL/min. The cut flow of STA decreased with aging (p=0.027) and increased with diameter (p=0.004). The cut flow showed no correlation with history of hypertension or diabetes mellitus (p=0.713 and p=0.786), but did correlate a positively with history of hyperlipidemia (p=0.004). There were no statistical differences in cut flow, STA diameter, and post anastomosis flow between the frontal and parietal branches (p=0.081, p=0.853, and p=0.990, respectively). Conclusion The median STA cut flow of a Korean population was 35 mL/min. Upon reviewing previous articles, it appears that there are differences in the STA cut flow between Western and Asian patients. INTRODUCTION Superficial temporal artery (STA)-middle cerebral artery (MCA) anastomosis is a surgical procedure performed as a countermeasure for hypoperfusion or parent vessel occlusion due to aneurysm trapping 3,11) .This is considered a low-flow bypass technique that utilizes a STA with a diameter of around 2 mm.After the STA is trimmed, the initial flow is very slow but gradually increases over time 1) .Manipulation of the vessel during harvesting can cause vasoconstriction, which can be relieved by applying papaverine and allowing the flow to increase slowly.In some cases, the STA may be damaged J Korean Neurosurg Soc 67 (2) : 158-165 during harvesting, in which case another donor vessel should be used for a successful bypass.Therefore, the surgeon should have a good understanding of the STA cut flow to determine if the blood flow is within a normal range.According to anecdotal evidence, the STA f low is between 15-25 mL/min among Western patients 6) , however, there is limited data available and no clear information about the Asian population.This study aims to introduce surgical techniques that maximize STA flow and to identify reference values for blood flow increase after cutting the distal end in Korean patients.The minimum and maximum cut flow values of the STA were measured directly during surgery. MATERIALS AND METHODS The Institutional Review Board of The Catholic University of Korea approved the current study (approval No. HC23RI-SI0019). A retrospective study was conducted on patients aged between 20 and 80 years who underwent STA-MCA anastomosis for intracranial atherosclerotic stenosis (ICAS) or Moyamoya disease (MMD) from May 2012 to October 2021.The patients were planned to undergo bypass surgery using preoperative magnetic resonance imaging (MRI), magnetic resonance angiography, perfusion MRI, single photon emission computed tomography (SPECT), and digital subtraction angiography (DSA).The indications for surgery were 1) acute cerebral infarction due to large cerebral artery occlusion, 2) a severe perfusion defect, e.g., with a T max 6 volume of 70 mL more and ineligibility for intra-arterial thrombectomy in ICAS patients and decreased reserve capacity from Diamox SPECT or a high risk of rebleeding in patients with MMD.The donor vessel was determined by DSA before surgery.In the case of a single bypass, the parietal branch of the STA was selected first.In the case of a double barrel, the frontal branch and parietal branch were harvested in turn. Operative method After assessing the subcutaneous course of the STA using Doppler imaging, skin marking and draping were completed, then, 0.5% lidocaine was injected into the perivascular galea to assist with STA dissection.A skin incision was made using a No. 15 surgical blade, and the vascularized galea flap was harvested using a Metzenbaum scissors.After harvesting the STA, the distal area of the vessel was cut and trimmed to match the anastomosis site. The cut flow was measured using a microvascular ultrasonic flowmeter (Charbel micro-flow probe; Transomics Systems, Inc., Ithaca, NY, USA), while the temporary clip of the proximal STA was opened.Immediately after harvesting the vessel, blood flow was ≤5 mL/min.At this time, cotton soaked with papaverine was applied to the blood vessel and the cut flow was measured at 5-minute intervals while waiting for resolution of the vasoconstriction that occurred during harvest.Measurements were repeated until they were increased and remained constant.If the change was ≤2 mL/min, it was recorded as the patient's final STA cut flow.Conversely, if the blood flow did not increase even after these attempts, further release of fibrous soft tissue covering the STA and papaverine administration were completed.The cut flow was measured at 5-minute intervals to increase the cut flow to ≤2 mL/min in the same way (Fig. 1).Thereafter, STA-MCA anastomosis was performed in the usual manner.After anastomosis, blood flow was measured through intraoperative micro-Doppler (Nicolet Biomedical, Madison, WI, USA) and patency was confirmed with an indocyanine green angiogram.Then, donor vessel flow was measured using a flowmeter and recorded as post-anastomosis flow.The ratio of cut flow to post anastomosis flow was calculated and recorded as the cut flow index (CFI).In cases with intraoperative CFI ≤0.5, an additional bypass procedure using another STA branch was performed for successful revascularization.The diameters of the donor and recipient vessels were measured using a micro-gauge.STA patency was confirmed through computed tomography (CT) angiography 6 hours after the surgery, and the final STA patency was evaluated by DSA after 6 months.Cases with good intracranial flow irrigation through the STA as confirmed by DSA were classified into the intact STA group, while those with occluded or weak intracranial flow were classified as the compromised STA group. Statistical method Patient characteristics were described using the median frequency and interquartile range.Linear regression was used to analyze the association between cut flow and patient factors.A generalized linear model was employed to compare the variables of the frontal and parietal branches that underwent https://doi.org/10.3340/jkns.2023.0125STA-MCA anastomosis.The Wilcoxon rank-sum test was applied to compare variables between the STA frontal branch and parietal branches, as well as to determine long-term STA patency at 6 months after anastomosis.The impact of hyperlipidemia was analyzed using analysis of variance.All p-values were 2-sided, and p-values ≤0.05 was considered significant.R software (version 4.1.3;R Foundation for Statistical Computing, Vienna, Austria) was used for all statistical analyses. RESULTS A total of 226 patients underwent STA-MCA anastomosis, when, including double-barrel cases, a total of 257 STAs was analyzed.The median age was 59.0 years (48.5-70.0).One hundred seventy-four patients had ICAS, while 52 had MMD (hemorrhagic or ischemic type).Totals of 94 frontal branches and 163 parietal branches were measured.In three cases, the flow was ≤10 mL/min and did not increase with any management.If donor vessel damage (e.g., intraluminal dissection of a partial tear) was confirmed, the damaged area was cut, and end-to-end anastomosis was performed to maintain flow.The cut flow, which was measured immediately after trimming the STA, was about 7-25 mL/min.One hundred twenty-nine cases showed a gradual increase in STA cut flow after applying papaverine.One hundred twenty-eight cases did not show an increase after applying papaverine alone and required release One hundred thirty cases showed a gradual increase in STA cut flow after application of papaverine (blue).One hundred twenty-eight cases required release of perivascular fibrous tissue surrounding the STA.Release of the perivascular soft tissue took an average of 10 minutes (dashed). of perivascular fibrous tissue surrounding the STA.Releasing perivascular fibrous tissue took an average of 10 minutes, and stable cut flow was maintained within 5-10 minutes after releasing the tissue (Fig. 2).The median cut flow of the whole case study population was 35.0 mL/min (6.0-98.0),while the median STA diameter was 1.8 mm (0.7-3.0), with median diameters of the frontal and parietal branches of 1.8, 2.0 mm, respectively.The median post-anastomosis flow measured using the flow meter was 24.0 mL/min (2.0-81.0).According to brain CT angiography performed 6 hours after the surgery, STA patency was intact in 254 cases, while flow was not identified in three cases.Of the total 257 bypasses performed in this study, follow-up angiography at 6 months confirmed persistency in 176 cases.In total, 157 bypass flows remained intact and were classified as the STA intact group (Table 1). Factors affecting cut flow were analyzed through multivariate analysis.As patient age increased, cut flow decreased significantly (p=0.027), as vessel diameter increased, cut flow increased significantly (p=0.004,Fig. 3).There was no significant difference in cut flow according to history of hypertension or diabetes mellitus (p=0.713 and p=0.786, respectively), but patients with hyperlipidemia presented higher cut flow values (p=0.004,Table 2).In the analysis conducted by dividing STA diameter using a 2 mm threshold, it has been observed that the presence of hyperlipidemia still inf luences the f low (p=0.013,Table 3). STA (n=257) Analysis was performed to assess the difference between the frontal branch and the parietal branch.The median cut flow of the parietal branch was 36 mL/min, while that of the frontal branch was 30 mL/min, showing no significant difference (p=0.081,Fig. 4).The median STA diameter was 1.8 mm in the frontal branch and 2.0 mm in the parietal branch, with no significant difference (p=0.853).Finally, there was no significant difference in post-anastomosis flow between the frontal and parietal branches (p=0.990)(Table 4). Further analysis of 157 cases with intact STA flow (intact group) and 19 cases with poor flow or occlusion (compromised group) on follow-up angiography was performed.The cut flow values displayed a non-significant difference between the groups, with median of 35.0 mL/min (7.1-85.0) in the intact group and 49.0 mL/min (16.4-98.0) in the compromised group (p=0.067).There was also no significant difference in diameter of STA, with a median of 1.8 mm (0.7-3.0) for the intact group versus that of 2.0 mm (1.0-2.0) in the compromised group (p=0.972).The median post-anastomosis flow was 24.0 mL/min in the intact group and 21.0 mL/min in the compromised group, with no statistical difference (p=0.364)(Table 5). Additional analyses for patients with MMD and ICAS were done.The cut flow between the two groups did not show a significant difference, with median values of 32.0 mL/min and 35.0 mL/min, respectively (p=0.774).However, the diameter was found to be significantly larger in the ICAS group (p<0.001).Regarding post anastomosis flow, no significant difference was observed between the two groups (p=0.076)(Table 6).For ICAS patients, we conducted an analysis of the CFI, which showed a median value of 0.64 (0.47-1.05) with a minimum of 0.06 and a maximum of 4.00.Regarding the long-term follow-up flow, CFI demonstrated borderline significance (p=0.063). DISCUSSION STA-MCA bypass is a low flow bypass procedure that is used for flow augmentation during intentional parent vessel occlusion for managing a large cerebral aneurysm or in cases of severe perfusion defects in patients with MMD or ICAS 3,10,11) .However, there are few quantitative investigations of the amount of blood supplied into the cranial cavity through an STA with about 2 mm in diameter.Cut flow is the blood flow measured through the dissected STA after cutting the distal end during anastomosis.Although it is not possible to confirm a exact value, low flow bypass refers to a flow ≤50 mL/min, and high flow bypass refers to a flow ≥50 mL/min 9) .In STA-MCA anastomosis, the flow is usually around 15-20 mL/min 6) .Conversely, in high flow graft, a flow of about 70-140 mL/min is supplied 8,13) .However, there is insufficient numerical data on these measures.According to our study, which investigated a Korean population, patients showed a median STA cut flow of 35 mL/min (6-98), and 18.6% of patients displayed a flow ≥50 mL/min.This analysis revealed that STA can supply more blood than expected. The relationship between cut flow and diameter according to Poiseuille's equation is that blood flow rate increases as the force of the vessel wall that causes resistance to blood flow in the blood vessel decreases 4) Therefore, the larger is the diameter, the higher is the cut flow (p=0.004).One could predict the flow through the larger diameter among the branches of STA, which indicates that a high cut flow can be obtained.Analysis of factors affecting cut flow showed that cut flow decreased with age (p=0.027).Neither a history of hypertension nor one of diabetes mellitus had a significant effect on STA cut flow (p=0.713 and p=0.786, respectively), but patients with hyperlipidemia had a higher cut flow than those without (p=0.004).Even considering effect of vessel diameter on cut flow, history of hyperlipidemia showed positive effect on cut f low (p=0.007).On the other hand, previous studies contend that hyperlipidemia is associated with a reduction in cerebral blood flow due to effects on blood viscosity or the vessel itself, and hyperlipidemia is an important factor in atherosclerosis 2,16) .The higher STA cut flow in patients with hyperlipidemia in this study suggest that hyperlipidemia has different tendencies in its effect in intracranial vessels and extracranial vessels, and additional research could be planned. When the difference between 94 frontal branches and 163 parietal branches was analyzed, the cut flow of the parietal branch showed no difference compared to that of the frontal branch (p=0.081).The median STA diameter in the frontal branch was 1.8 mm and that in the parietal branch was 2.0 mm, while the median post anastomosis flow was 24.0 mL/min in the frontal branch and 23.80 mL/min in the parietal branch, without a significant difference (p=0.990). However, it was confirmed by 6-month follow-up angiography that there is no significant difference in STA cut flow between the intact and compromised group (p=0.067), and post anastomosis flow showed no difference between these groups In five cases, the flow through the STA was weak, but the collateral flow through the posterior cerebral artery or anterior cerebral artery was increased at long-term follow up angiography.An absolute comparison between patients was not possible because samples could not be extracted under the same conditions.However, comparison of the cut flow values of Western studies with those of a study from China and this study suggests that the cut flow values in Eastern studies tend to be lower than those of Western populations 1,12,15) (Table 7).There was a scatterplot of Western patients in the article of Amin-Hanjani et al. 1) and the scatterplot of Korean patients was made under our data in the same format.Highest cut flow is significantly low in our data (Fig. 3).In a study analyzing the effect of ethnicity on the diameter of the carotid artery, the Hispanic group had a significantly smaller artery diameter but greater stiffness than the Caucasian group 7) , and it is reasonable that there would be ethnic differences in cut flow, diameter, and post anastomosis flow of the STA.Also, there were differences in inclusion criteria between Western and Asian studies.Western studies only included patients with ischemic disease 1,12) , but the study from China and our study included patients with ischemic disease or MMD 15) .Previous studies have analyzed differences in clinical features and incidence rates of MMD according to ethnicity 5,14) .These ethnic differences may affect not only the occurrence of MMD, but also, the clinical features of cranial arteries, including the STA. In institutions without ultrasonic flowmeters for measuring cut flow, numerical evaluation of STA cut flow is possible by measuring the amount of blood dropped into a bottle for 1 minute after trimming the STA.This method enables a more quantitative approach to STA-MCA anastomosis and will help improve the surgical outcome. CONCLUSION This study analyzed STA flow, post anastomosis flow, and CFI in Korean patients using a total of 257 STAs that underwent STA-MCA bypass.The median STA cut flow of Koreans was 35 mL/min.After reviewing previous articles, it seems there are differences in the STA cut flow observed between Western and Asian patients. Fig. 1 . Fig. 1.Operative method of superficial temporal artery (STA)-middle cerebral artery (McA) anastomosis.A : The vascularized galeal flap was harvested using a Metzenbaum scissors.b : Immediately after harvest, blood flow is low.c : Removing fibrous soft tissue covering the STA.d : After removing fibrous soft tissue covering the STA and applying papaverine, the cut flow increased gradually.e : Using microvascular flow-meter, cut flow of STA was measured.f : end-to-side anastomosis of STA-McA was performed. Fig. 2 . Fig. 2.The cut flow value of superficial temporal artery (STA) over time.One hundred thirty cases showed a gradual increase in STA cut flow after application of papaverine (blue).One hundred twenty-eight cases required release of perivascular fibrous tissue surrounding the STA.Release of the perivascular soft tissue took an average of 10 minutes (dashed). Female 5 )Fig. 3 . Fig. 3. Scatterplot of cut flow according to diameter of superficial temporal artery. Fig. 4 . Fig. 4. The cut flow of the frontal and parietal branches. Table 2 . Multivariate analysis for the cut flow of the superficial temporal artery Table 3 . Analysis of diameter differences in relation to the presence of hyperlipidemia Table 4 . comparative analysis between a frontal branch and parietal branch Table 5 . Multivariate analysis for the flow of follow up angiography Table 7 . STA cut flow of other studies in bypass surgery p=0.364).If post-anastomosis flow is maintained, it can be interpreted as securing long-term patency.In 14 of 19 poor or occluded cases, recanalization of the parent vessel occurred.	2023-09-09T06:17:42.944Z	2023-09-07T00:00:00.000	{ "year": 2023, "sha1": "353f186f38ac652df35c21db697eacdd22742b4d", "oa_license": "CCBYNC", "oa_url": "https://www.jkns.or.kr/upload/pdf/jkns-2023-0125.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "d91cdefb7d2ec17fcd45cd6b77c289d8b7b02c12", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
20225771	pes2o/s2orc	v3-fos-license	Saving Lives and Saving Money A new Medicaid system is emerging in North Carolina in which accountable care organizations will aim to improve both the quality and value of health care. We explore how local health departments can apply their expertise in population health to help achieve these goals. A new Medicaid system is emerging in North Carolina in which accountable care organizations will aim to improve both the quality and value of health care. We explore how local health departments can apply their expertise in population health to help achieve these goals. T his article uses the culture of health framework developed by the Robert Wood Johnson Foundation to address the critical contributions local health departments can provide in a Medicaid managed care environment. This framework identifies 5 strategies for improving population health: partnerships, innovation, data, connections, and payment reform. Improving population health is one of the stated goals of Medicaid reform in North Carolina. Partnerships Between Public Health and Health Care Partnerships between public health and health care will be critical in a Medicaid managed care environment in order to ensure a comprehensive view of factors affecting health outside the scope of traditional clinical care. Health behaviors, physical environment, and socioeconomic factors constitute up to 80% of the outcome equation in many population health models [1]. These nonclinical aspects are even more important when considering the vulnerable populations that are eligible for Medicaid. One of the fundamental obligations of public health is to perform a community health assessment that evaluates the factors that influence health and quality of life. Tax-exempt hospitals are also required to conduct a community health needs assessment to assure that they have the information they need to prioritize and coordinate their community benefits with other health improvement efforts. Although hospitals and health departments previously pseudopartnered on assessments, North Carolina recognized the potential synergy for these 2 processes and promoted connectivity through the North Carolina Public Health/ Hospital Collaborative to facilitate and advance shared health improvement initiatives through public health and hospital partnerships [2]. There are other public health and health care partnerships that are structured for managed care environments such as population care management programs for children enrolled in Medicaid and pregnant women at elevated risk of complications. These programs are a shared responsibility of regional Community Care of North Carolina networks, local health departments, and local medical practices. Public health nurses and social workers are embedded in pediatric and obstetric/gynecology practices to connect patients and families with needed services and resources outside the scope of clinical care; these services can include transportation, medication management, referrals, home visits, and health education. Health departments provide multiple wraparound services in this domain-with supports like nutrition counseling, immunizations, family planning, breastfeeding consultation, and smoking cessation-while maintaining coordination with the medical home. Traditional medical management programs show limited effectiveness [3][4][5] given their focus on specific disease conditions and their lack of linkage to preventive or population care. Medical management combined with population care management is increasingly viewed as a promising intervention and best practice in improving health outcomes [6,7]. Innovation to Address Drivers of Health Drivers of health require more than a technical fix; they require adaptive and innovative responses to health on a physical, social, emotional, and fiscal front. Changes of this magnitude require a collective approach. The understanding that health is more than just the absence of illness comes to the forefront of this discussion, particularly as recognition of the importance of social determinants of health collides with payment reform processes. For example, while specific population care management programs may provide muchneeded medication through an assistance program or a crib to ensure a safe sleep environment for a new baby, these programs are rarely robust enough to also tackle the greater needs of safe and stable housing, steady and livable income, or quality education. Health care reform provides health departments, hospitals, and other community partners an opportunity to build on community health assessment partnerships by collaborating on collective impact approaches. In a collective impact approach, organizations' individual agendas are set aside in favor of pursuing a shared and aligned agenda around specific outcomes. These specific outcomes are typically broader in scope and are also more likely to move the needle in terms of health and quality of life for a population. This approach allows local health departments to be the social determinant navigator for accountable care organizations. As hospitals embrace their role in improving community health, there is an opportunity for health departments to be the community "hand" that leads an individual out of the hospital and into an environment that will promote and support clinically relevant behavior changes. These opportunities may include environmental or systems changes like increased access to healthy foods through EBT/SNAP utilization at farmers' markets or rules and ordinances that maintain tobacco-free public places. Looking at local policy changes through a "health in all policies" lens further supports a patient's individual medical management and also integrates the population-based approach supported by public health. New Ways to Use and Share Data, Analytics, and Information Technology Health care reform provides an opportunity for public health and hospitals to build their partnerships and work in new ways. One avenue is in data and information technology. Local health departments can be the navigators as hospitals enter the world of collective impact. Collective impact involves the establishment of shared goals around specific data indicators, many of which local health departments already track and monitor. Other indicators require the sharing of data from a variety of sources such as hospital systems, education systems, and mental health providers. Medicaid reform with a focus on cost savings and quality of care could facilitate collaboration across sectors, including data sharing. Additionally, as public health departments and hospitals explore integration of medical and population care management, continuity of electronic medical record systems becomes important for the sharing of patient information, population trends, and linkages of community resources. Ideally, if using the same systems, both the hospital and the county health department have access to the patient's medical treatment plan and to his or her community coordinated care action plan. Knowledge of both plans by each entity enhances the continuity of care, thereby improving the quality of that care. Further, hospitals can utilize the talents of local health departments by allowing access to admissions data that can then be mapped using a geographic information sys-tem (GIS). With this data, public health departments can map hot spots of chronic disease or other health issues. GIS mapping can then aid in a population-based management approach by targeting culturally specific interventions in neighborhoods or potentially by identifying gaps in services or resources based on the location of clusters. Further, health departments and hospitals can potentially synergize efforts in order to create regional data hubs that can collect the locally specific data critical in addressing prioritized social determinants of health. This is of particular importance as the availability of local morbidity and behaviorrelated data for counties has whittled away as changes in and reduced funding for the Behavioral Risk Factor Surveillance Survey have been implemented. Connecting Patients to Social and Community-Based Services Public health has traditionally focused on a broad array of linkages and referrals to support services for their patients and external clients. This expertise is rooted in our history of care management services, but that orientation has also been influenced by the reality that many social determinants are addressed by other community organizations. Since health departments are principal safety-net providers in many communities and often deal with complex circumstances related to their patients, they are frequently in the position of providing resource connections for these populations. It is a responsibility born of necessity: If patients lack transportation to receive regular medical care or lack food to meet basic nutrition needs, then it is unrealistic to expect improvements in health until these needs are addressed. Accordingly, health departments have invested in a variety of different assets including interpreters, social workers, outreach and education specialists, community health workers, navigators, prescription assistance aides, and behavioral health counselors. These staffing resources are driven by local needs and community support. In addition to internal capacity, public health departments have also embedded supplemental operations within their departments by including staff from social services, mental health services, federally qualified health centers, free clinics, and other human services agencies. These quasi-formal systems are extensive and are consistent with academic characterizations of an integrated delivery system as "a network of organizations that provides or arranges to provide a coordinated continuum of services to a defined population" [8]. These components are precursors to accountable care or managed care structures, and further harnessing the inherent benefit of these delivery systems will provide mutual benefits to populations, managed care organizations, and public health. Emerging Payment Models to Incentivize and Sustain Initiatives Both the Patient Protection and Affordable Care Act of 2010 and the Medicare Access and Children's Health Insurance Program Reauthorization Act of 2015 have reinvigorated the focus on value and quality in health care payment. Value and quality in health care outcomes require a holistic view of the patient and their environment. As previously illustrated, local health departments-with their presence in all 100 North Carolina counties and their focus on integrated systems and services-are uniquely positioned to provide this view, as well as the community-based interventions necessary to assure a healthier population. How then, are these services financially supported in Medicaid reform? There is emerging work around risk-adjusting payments, which involves paying more for patients deemed to be at higher risk for adverse outcomes. This work supports the proposition that medical providers like health departments should be paid more to treat patients with greater needs, such as patients living in poverty, those with lower educational status, or individuals who face transportation or language barriers. However, paying more to treat patients with greater needs seems like a baby step when what is needed is a lunar jump. To truly address these health-impeding, complex issues requires investment upstream to improve social determinants of health. In fact, it is possible that these riskadjusted models are slowing the transition to a more comprehensive solution. Where is the incentive to help people lead healthier lives if payment is greater for those who lead less healthy lives? Relatedly, compelling research being done by ReThink Health's dynamics model shows that, for long-term savings in lives and costs, the most effective strategy is a reform that addresses health care quality and coverage as well as invests in population-based strategies to ensure healthier living. Specifically, they found that when added to quality and coverage, population health interventions would save 90% more lives and reduce costs by 30% after 10 years. At 25 years, those numbers improve, with population health interventions saving 140% more lives and reducing costs by 62% [9]. Fully funding tobacco, diabetes, and other chronic disease prevention programs, for example, and then (and here's our lunar jump) reinvesting initial savings generated from reform back into the system yields the kind of results about which every health system reformer dreams. Fortunately, North Carolina's proposed Medicaid reform plan includes this concept of reinvesting savings in the form of a delivery system reform incentive payment (DSRIP). While limited in scope, DSRIP does allow local health departments to start the work necessary to demonstrate that an investment in population health saves lives and money. Conclusion A commitment to the broader model of population health that addresses the social determinants of health, paired with innovative investments generated by Medicaid reform efficiencies, will improve the quality and value of health care in North Carolina. Capitalizing on and fully funding the population health infrastructure already in place through our local health departments is an effective strategy to accomplish Medicaid managed care organizations' goals of improving health and reducing costs.	2018-04-03T02:06:05.921Z	2017-01-01T00:00:00.000	{ "year": 2017, "sha1": "4a86af8bb45619e56147718e308c0f56baa29a5a", "oa_license": null, "oa_url": "http://www.ncmedicaljournal.com/content/78/1/55.full.pdf", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "ac11cd5640b94ac4b7b72a92ebb97baf15360679", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
232103113	pes2o/s2orc	v3-fos-license	Current Insights into 3D Bioprinting: An Advanced Approach for Eye Tissue Regeneration Three-dimensional (3D) printing is a game changer technology that holds great promise for a wide variety of biomedical applications, including ophthalmology. Through this emerging technique, specific eye tissues can be custom-fabricated in a flexible and automated way, incorporating different cell types and biomaterials in precise anatomical 3D geometries. However, and despite the great progress and possibilities generated in recent years, there are still challenges to overcome that jeopardize its clinical application in regular practice. The main goal of this review is to provide an in-depth understanding of the current status and implementation of 3D bioprinting technology in the ophthalmology field in order to manufacture relevant tissues such as cornea, retina and conjunctiva. Special attention is paid to the description of the most commonly employed bioprinting methods, and the most relevant eye tissue engineering studies performed by 3D bioprinting technology at preclinical level. In addition, other relevant issues related to use of 3D bioprinting for ocular drug delivery, as well as both ethical and regulatory aspects, are analyzed. Through this review, we aim to raise awareness among the research community and report recent advances and future directions in order to apply this advanced therapy in the eye tissue regeneration field. Introduction The development of new revolutionary technologies during recent years, such as the use of Big Data, virtual reality systems and three-dimensional (3D) bioprinting, has created great expectations in the scientific community, not only regarding the improvement of the quality of life in patients affected by devastating pathologies, but also in terms of saving health-care associated resources [1][2][3]. In this regard, 3D bioprinting is an emerging manufacturing technology which holds great promise for a wide variety of biomedical applications, including drug testing, pathophysiological studies and regenerative medicine [4]. Specific benefits of such a technology would include the development of more targeted and personalized therapeutic approaches, as well as the possibility of obtaining functional models of tissues and organs for research purposes, increased reproducibility and higher capacity for drug lab testing studies, etc. In addition, for regenerative medicine applications, key advantages include automated tissue fabrication and the flexibility of incorporating many different materials and cell types in precise anatomical 3D geometries [4]. Consequently, the interest and investment in this promising technology has From a conceptual standpoint, 3D bioprinting refers to the "additive manufacturing" process based on a layer-by-layer approach with the deposition of bio-inks in a precise spatial arrangement [6], which makes it a suitable technology for obtaining highly complex structures similar to the original tissues of the eye [7]. The bio-ink is composed of structural components or biomaterials, crosslinkers, functional elements and living cells. Compared to non-biological printing, which has been proved suitable for the manufacturing of medical devices and patient-tailored prosthetics for more than 30 years [8], 3D bioprinting involves additional challenges and requires the multidisciplinary integration of different technological and medical fields. These complexities refer mainly to the selection of biocompatible materials, cell types, biomechanical cues and the overcoming of technical difficulties due to the sensitivities of living cells [9]. By using different materials and designs, the structural, physicochemical and mechanical properties of the bio-printed structure can be adjusted [10]. In general terms, the fundamental aspects that need to be taken into account for an appropriate selection of biomaterials include their printability, biocompatibility, degradation kinetics and byproducts, structural and mechanical properties, and biomimicry. Nowadays, for eye tissue regeneration applications, the most commonly used biomaterials in bio-inks are based on naturally derived polymers such as collagen, gelatin, alginate or hyaluronic acid [11], whose similarity to human extracellular matrix (ECM) and inherent bioactivity represent a relevant advantage [9]. Additional to the selection of suitable biomaterials, the target tissue determines the type of cells that need to be used in the bio-ink. In most cases, the tissues of the eye are made up of more than one cell type. For this reason, many of the approaches made to date incorporate more than one cell type or contemplate, as future steps, the incorporation of more cell types in the obtained scaffolds. In 3D bioprinting, the three major techniques include inkjet bioprinting, based on the deposition of bio-ink droplets, laser-assisted bioprinting, based on laser stimulation, and extrusion bioprinting, which uses mechanical force to deposit a continuous flow of bio-ink, each presenting specific features [4]. The choice of the most suitable approach depends on the specific application of interest, since each 3D printing technique holds its own peculiarities and these result in different outcomes [12][13][14][15][16][17] (Figure 2). From a conceptual standpoint, 3D bioprinting refers to the "additive manufacturing" process based on a layer-by-layer approach with the deposition of bio-inks in a precise spatial arrangement [6], which makes it a suitable technology for obtaining highly complex structures similar to the original tissues of the eye [7]. The bio-ink is composed of structural components or biomaterials, crosslinkers, functional elements and living cells. Compared to non-biological printing, which has been proved suitable for the manufacturing of medical devices and patient-tailored prosthetics for more than 30 years [8], 3D bioprinting involves additional challenges and requires the multidisciplinary integration of different technological and medical fields. These complexities refer mainly to the selection of biocompatible materials, cell types, biomechanical cues and the overcoming of technical difficulties due to the sensitivities of living cells [9]. By using different materials and designs, the structural, physicochemical and mechanical properties of the bio-printed structure can be adjusted [10]. In general terms, the fundamental aspects that need to be taken into account for an appropriate selection of biomaterials include their printability, biocompatibility, degradation kinetics and byproducts, structural and mechanical properties, and biomimicry. Nowadays, for eye tissue regeneration applications, the most commonly used biomaterials in bio-inks are based on naturally derived polymers such as collagen, gelatin, alginate or hyaluronic acid [11], whose similarity to human extracellular matrix (ECM) and inherent bioactivity represent a relevant advantage [9]. Additional to the selection of suitable biomaterials, the target tissue determines the type of cells that need to be used in the bio-ink. In most cases, the tissues of the eye are made up of more than one cell type. For this reason, many of the approaches made to date incorporate more than one cell type or contemplate, as future steps, the incorporation of more cell types in the obtained scaffolds. In 3D bioprinting, the three major techniques include inkjet bioprinting, based on the deposition of bio-ink droplets, laser-assisted bioprinting, based on laser stimulation, and extrusion bioprinting, which uses mechanical force to deposit a continuous flow of bio-ink, each presenting specific features [4]. The choice of the most suitable approach depends on the specific application of interest, since each 3D printing technique holds its own peculiarities and these result in different outcomes [12][13][14][15][16][17] (Figure 2). Some success has been demonstrated in early attempts to recreate complex tissue structures, and latest research evidences significant improvement in terms of effectiveness, resolution, accuracy and manufacturing speed of this customizable technique. All of this suggests, from a technical point of view, that the fabrication of biocompatible and functionalized full bio-printed organs will be possible in the near future [18]. In this regard, the success of 3D bioprinting for tissues and organs also depends on the target organ or tissue. For instance, for structurally simple tissues such as skin, 3D bioprinting is close to becoming a clinically relevant method for producing skin grafts and is already being Pharmaceutics 2021, 13, 308 3 of 28 used in the cosmetic industry [4]. On the other hand, for more complex organs such as the heart, it becomes more challenging to precisely reproduce its diverse functionality and heterogeneous composition, so is still far from clinical translation [4]. Some success has been demonstrated in early attempts to recreate complex tissue structures, and latest research evidences significant improvement in terms of effectiveness, resolution, accuracy and manufacturing speed of this customizable technique. All of this suggests, from a technical point of view, that the fabrication of biocompatible and functionalized full bio-printed organs will be possible in the near future [18]. In this regard, the success of 3D bioprinting for tissues and organs also depends on the target organ or tissue. For instance, for structurally simple tissues such as skin, 3D bioprinting is close to becoming a clinically relevant method for producing skin grafts and is already being used in the cosmetic industry [4]. On the other hand, for more complex organs such as the heart, it becomes more challenging to precisely reproduce its diverse functionality and heterogeneous composition, so is still far from clinical translation [4]. The field of ophthalmological applications of 3D bioprinting is also gaining interest due to the specific features of the eye ( Figure 1B). In addition, its anatomical disposition provides easy access for surgery and implantation in both the inner layer and overall in the anterior chamber. Furthermore, the privileged immune condition of the eye coupled with the disposition of multiple diagnostic tools, turn it into an attractive organ for the application of 3D bioprinting technology, reducing the lack of organ donors along with the problems associated with rejection of transplanted grafts due to inappropriate immune response [19]. Some of the ophthalmological applications are the development of 3D bioprinting of anatomically realistic ocular models to enhance both education and clinical practice, providing better training opportunities, or the design of cost-effective personalized approaches by manufacturing specific ocular structures to treat serious eye diseases that affect more than 30% of the world's population [20]. Depending on the The field of ophthalmological applications of 3D bioprinting is also gaining interest due to the specific features of the eye ( Figure 1B). In addition, its anatomical disposition provides easy access for surgery and implantation in both the inner layer and overall in the anterior chamber. Furthermore, the privileged immune condition of the eye coupled with the disposition of multiple diagnostic tools, turn it into an attractive organ for the application of 3D bioprinting technology, reducing the lack of organ donors along with the problems associated with rejection of transplanted grafts due to inappropriate immune response [19]. Some of the ophthalmological applications are the development of 3D bioprinting of anatomically realistic ocular models to enhance both education and clinical practice, providing better training opportunities, or the design of cost-effective personalized approaches by manufacturing specific ocular structures to treat serious eye diseases that affect more than 30% of the world's population [20]. Depending on the specific eye disease to be treated, different ocular structures can be 3D bio-printed taking into account each disease's specific structural and functional complexities. As known, the eye is a complex, isolated, highly evolved organ with different and specifically organized tissue structures, which embrace, from an anatomical point of view, "simple" cell multilayer structures such as the cornea and more complex structures enclosing central nervous system cells, such as the retina. However, it is noteworthy that each eye tissue possesses its complexities when implementing 3D bioprinting. In this regard, the cornea represents a key ocular transparent tissue for vision that can be Pharmaceutics 2021, 13, 308 4 of 28 bio-printed, due its relatively simple structural disposition, into five independent layers. In normal conditions, the different hydrophilic and hydrophobic permeability of such layers, along with the interconnected tight junctions in epithelial cells, can severely limit the access of external agents into inner ocular structures, acting as a potent biological barrier [21]. However, severe alterations in the corneal thickness along with keratitis of multiple causes can lead to advanced stages of corneal tissue damage, where corneal transplant is the last medical choice. In this scenario, 3D bio-printed corneal tissue could emerge as an interesting medical option, bearing in mind that severe corneal disorders are the main cause of blindness worldwide [20]. In addition, it is worth mentioning that nowadays there is a lack of full-structure corneal in vitro models for both drug screening and toxicological testing, and that the deficit of donated corneas has increased the urgency of transplantable corneal substitutes. On the other hand, there are other relevant ocular tissues such as the retina, where the application of 3D bioprinting technology is more challenging from a scientific point of view, due to its complexity, neurological origin and connections with other areas of the central nervous system [22]. Some retinal dystrophies that dramatically hamper human life and for which there are no currently effective treatments include retinitis pigmentosa, Stargardt disease and age-related macular degeneration, to name just the most representative. Therefore, the possibility of bio-printing functional retinal tissue for both drug testing and graft implantation purposes has recently caught the attention of the scientific community in order to offer an alternative medical approach against such devastating retinal pathologies. In any case, the field of 3D bioprinting for regenerative medicine is in its early stages and there are still several technological challenges to overcome, as well as some relevant ethical and regulatory concerns to be addressed, before fabricating large scale organs of all levels of complexity [23]. For instance, to become a realistic medical option, several parameters still need to be deeply considered such as the biocompatibility and mechanical properties of engrafts along with the biological behavior and attachment properties of cells, to name just the most relevant [22]. Nevertheless, if improvement and evolution in 3D bioprinting technology continues to progress exponentially, the creation of artificial biosynthetic and customizable full eyeballs in the near future is likely, along with other applications that have not yet been fathomed [24,25]. This review further discusses the current status and implementation of 3D bioprinting technology in the ophthalmology field in order to manufacture relevant tissue-engineering items such as the cornea, retina and conjunctiva. Special attention will be paid to the description of the most commonly employed bioprinting methods, and the most relevant eye tissue engineering studies performed by 3D bioprinting technology at preclinical level. In addition, regulatory, ethical and future directions related to the use of this "gamechanger" technology in ophthalmology will be addressed. 3D Bioprinting for Eye Tissue Engineering The eye is a very complex organ formed by different structures within which are the orbit, the sclera, the conjunctiva, the cornea, the iris, the pupil, the lens, the vitreous humor, the retina and the optic nerve [26]. Generally, each of the diseases that occur in the eyes affect only one of the aforementioned structures [27]. In this sense, the approximations carried out to date using 3D bioprinting have as their main objective the production of structures that resemble the characteristics and properties of the affected tissue. Such 3D-structures would allow the replacement of the entire damaged tissue or a part of it, in order to restore the patient's vision. So far, the three main target tissues have been the cornea, the retina and the conjunctiva. Cornea The cornea is an innervated and avascular tissue located in front of the pupil and iris [28]. Its main characteristic is its transparency, which allows the transmittance and refraction of the light entering the eye [20]. It acts as a mechanical and chemical barrier Pharmaceutics 2021, 13, 308 5 of 28 protecting the inner eye from external agents such as mechanical damage, microorganisms or ultraviolet radiation [29]. As a complex tissue, it is divided into five differentiated layers ( Figure 3): epithelium, Bowman's membrane, stroma, Descement's membrane and endothelium [28,30]. order to restore the patient's vision. So far, the three main target tissues have been the cornea, the retina and the conjunctiva. Cornea The cornea is an innervated and avascular tissue located in front of the pupil and iris [28]. Its main characteristic is its transparency, which allows the transmittance and refraction of the light entering the eye [20]. It acts as a mechanical and chemical barrier protecting the inner eye from external agents such as mechanical damage, microorganisms or ultraviolet radiation [29]. As a complex tissue, it is divided into five differentiated layers ( Figure 3): epithelium, Bowman's membrane, stroma, Descement's membrane and endothelium [28,30]. The corneal epithelium is composed of a few layers of epithelial cells that form the outermost area of the cornea. The epithelial cells are constantly renewed from the basal layer, which is formed by limbal stem cells (LSCs) [31,32]. However, any damage in the basal area would lead to a dysfunction of the LSCs that would result in overgrowth of the conjunctiva and blood vessels, photophobia and pain [32]. The Bowman's membrane is a thin acellular layer that separates the epithelium from the stroma [31]. The stroma is the widest part of the cornea. It is composed of laminin and collagen I fibrils that align perfectly to form a complex structure. This structural complexity is the key to its transparency and mechanical resistance [33,34]. In addition, it is composed of keratinocyte cells, which are responsible for maintaining the ECM [31]. These cells have little mitotic activity and present dendritic morphology. Nevertheless, in case of trauma, they are activated as a fibroblast that can differentiate into myofibroblasts [34,35]. Myofibroblasts express proteins that can alter the ECM causing opacity of the cornea, contraction and a corneal scar formation [31,35]. Descement's membrane is an acellular layer that separates the stroma from the endothelium [31]. The endothelium is the deepest layer and is composed of endothelial cells that are responsible for maintaining the fluid balance of the cornea [31]. They have very little capacity of regeneration in vivo, therefore any damage in this area would cause irreversible blindness [36]. Taking into account the structural complexity of the cornea, degeneration of any of its parts could lead to serious diseases. In fact, bilateral blindness due to corneal damage has a high prevalence worldwide [29] and it is estimated that more than four million The corneal epithelium is composed of a few layers of epithelial cells that form the outermost area of the cornea. The epithelial cells are constantly renewed from the basal layer, which is formed by limbal stem cells (LSCs) [31,32]. However, any damage in the basal area would lead to a dysfunction of the LSCs that would result in overgrowth of the conjunctiva and blood vessels, photophobia and pain [32]. The Bowman's membrane is a thin acellular layer that separates the epithelium from the stroma [31]. The stroma is the widest part of the cornea. It is composed of laminin and collagen I fibrils that align perfectly to form a complex structure. This structural complexity is the key to its transparency and mechanical resistance [33,34]. In addition, it is composed of keratinocyte cells, which are responsible for maintaining the ECM [31]. These cells have little mitotic activity and present dendritic morphology. Nevertheless, in case of trauma, they are activated as a fibroblast that can differentiate into myofibroblasts [34,35]. Myofibroblasts express proteins that can alter the ECM causing opacity of the cornea, contraction and a corneal scar formation [31,35]. Descement's membrane is an acellular layer that separates the stroma from the endothelium [31]. The endothelium is the deepest layer and is composed of endothelial cells that are responsible for maintaining the fluid balance of the cornea [31]. They have very little capacity of regeneration in vivo, therefore any damage in this area would cause irreversible blindness [36]. Taking into account the structural complexity of the cornea, degeneration of any of its parts could lead to serious diseases. In fact, bilateral blindness due to corneal damage has a high prevalence worldwide [29] and it is estimated that more than four million people suffer from a corneal disease [35]. With this demand, keratoplasty or corneal transplantation is the treatment of choice. However, the possibility of rejection, the poor survival of the explanted tissue, the absence of corneal banks, the high cost and scarce accessibility to transplantation necessitate the consideration alternative treatments [29,37]. Cell therapy [38] and the manufacturing of structures such as acellular membranes [39] have been proposed in order to supply high transplantation demand. However, it has been impossible to obtain a fully functional corneal substitute. In this context, 3D bioprinting has emerged as a promising technology since complex multilayer tissues, such as the cornea, can be easily reproduced. Furthermore, this technique can be advantageous over other techniques, since scaffolds with characteristics similar to those of the native cornea can be obtained. These characteristics are listed below. Transparency. Transparency is its greatest characteristic. Previously proposed manufactured therapies, such as decellularized membranes, have had difficulties in achieving a transparency similar to that of the native cornea. As the key to transparency resides in a perfectly arranged structure, the use of 3D bioprinting that deposits layers exactly, as a pre-designed form, could be the solution. Biomechanics. The biomechanical properties of the cornea affect corneal curvature, strength, and conformability [40]. Previously proposed treatments, such as cell therapy, have paid little attention to these parameters. The mechanical strength can be adjusted by combining different biomaterials. Conventional treatments, such as membranes, use a single material, thereby limiting the control of the biomechanical properties. On the contrary, 3D bioprinting is a technology that allows the use of a great diversity of materials with very diverse properties. In this way, the mechanical properties can be adjusted to the needs of the corneal tissue. Curvature. Optical parameters of the cornea, such as light refraction, are due to its curvature [41]. Based on corneal geometrical information and using computer designed programs, a corneal prototype can be perfectly fabricated. This could be easily carried into the bioprinter achieving what would be hard to obtain with conventional techniques. Multilayer structure. The cornea is a complex multilayer tissue. It is composed of different layers in which different materials, cells and internal structure are found. Conventional cell therapies have been shown ineffective due to the poor survival rate and functionality of the implanted cells [36]. Furthermore, single material membranes have not achieved the multilayer complexity of the cornea. Thus, 3D bioprinting overcomes this problem as it is based on the deposition of materials layer by layer. In addition, different cell types (epithelial, keratocytes and endothelial cells) can be embedded into biocompatible biomaterials increasing cell viability and functionality. As a result, a corneal native-mimicking structure with the five differentiated layers could be performed. In this context, many studies have been carried out using 3D bioprinting technology for corneal tissue fabrication. Overall, these research studies have focused on the main part of the cornea, the stroma. Isaacson et al. [42] used extrusion based bioprinting for the development of a structure mimicking the stroma by embedding human keratocyte cells into an alginate/methacrylate type I collagen ink. They proposed the application of a rotational Scheimpflug camera in order to design patient-specific corneal model. Likewise, they developed a supportive structure to maintain the scaffold curvature during the bioprinting procedure. Consequently, they were able to reproduce the curved corneal geometry ( Figure 4A-a) and cell viability was high for 7 days after bioprinting ( Figure 4A-b). However, properties of great importance such as transparency and mechanical properties were not mentioned. These parameters were taken into a mixture of cells and ink placed onto slabs as a control. The study demonstrated that transparency and mechanical properties of bio-printed scaffolds were higher than those in the slabs, and data were similar to the native human cornea. Nevertheless, when analysing cellular studies, results were not as promising. Although cell viability was high in both systems, in both bio-printed scaffold and in slabs, cells were elongated and showed a dendritic morphology associated with keratocytes ( Figure 4B-b). Therefore, cells' metabolic activity and protein expression was low. It is believed that the high crosslinking density of 3D bio-printed scaffolds together with the absence of a curve geometry could negatively affect cell behaviour. Kim et al. [43] studied this abnormal cell behaviour exhaustively after extrusion bioprinting. They wanted to analyze how the shear stress, applied when extruding through the printing nozzle, affects cell behaviour. To do so, they proposed to bio-print, by using different nozzle diameters, human keratocytes into a bio-ink based on decellularized corneal ECM in order to reproduce corneal stroma. Results demonstrated that not only shear stress affected cell behaviour, but also the deposition of collagen fibrils. In fact, by bioprinting with wide nozzle diameters, no aligned collagen fibrils were observed in scaffolds, Pharmaceutics 2021, 13, 308 7 of 28 which decreased transparency. Furthermore, cell dendritic morphology and keratocyte specific gene expression were not found. In contrast, after bioprinting using narrower nozzles, the shear stress increased and, as a result, highly structured collagen fibrils were shown. Nevertheless, cells were damaged and showed fibroblastic behaviour. The authors concluded that with the application of proper force, they could achieve scaffolds with both characteristics: structured collagen fibrils and cells with keratocyte behaviour. Thus, they bio-printed scaffolds with the proper extrusion nozzle (25G) and, after demonstrating that they met the adequate characteristics, scaffolds were implanted into rabbits. In vivo studies showed that implanted scaffolds were optically more transparent than the control (the not printed implant). In addition, keratocytes' cellular behaviour was activated, which enhanced collagen production simulating a lattice pattern similar to the structure of native human cornea stroma [43]. harmaceutics 2021, 13, x geometry, curvature and thickness of the obtained scaffold were more similar to the native human cornea than those obtained in the other studies carried out contrast, extrusion bio-printed layers showed high diversity among algina bio-ink mixtures developed in this study, in terms of printability and mechani erties, even though, the overall transparency was good. Cell viability was high b authors focused mainly on improving the manufacturing precision and ge control, more in vitro studies should be carried out prior to taking these scaffol vivo studies. The previous studies made an approach towards manufacturing differen the cornea with different materials and techniques. However, to date, the only which the bioprinting of two different cellular parts of the cornea has been tried Sorkio et al. [32]. They proposed the development of a corneal epithelium and In order to avoid keratocyte stress through the nozzle when extrusion bioprinting is used, Duarte et al. [37] proposed another approach; the use of the inkjet bioprinting technique. This study focused on bioprinting a stroma-mimicking structure by using type I collagen/agarose and human keratocytes bio-ink. After bioprinting, biomechanical prop-Pharmaceutics 2021, 13, 308 8 of 28 erties, transparency, and cell viability and behaviour were analyzed. Results showed good printability using inkjet bioprinting, achieving curved and transparent scaffolds ( Figure 4C-a). Moreover, cell viability was high. Rounded morphology was observed at day 1 after bioprinting. Nevertheless, cells became dendritic and showed keratocyte phenotype after 7 days ( Figure 4C-b). In contrast, mechanical strength was lower than that of the human native cornea, so, improving mechanical properties would be their next main objective. When the objective is to bio-print the corneal epithelium, fewer studies have been published so far. Wu et al. [45] proposed to bio-print human corneal epithelial cells embedded into alginate/gelatin ink, in which different concentrations of collagen I were added. Extrusion bioprinting was the technique that they selected in order to bio-print square grid-like scaffolds ( Figure 5A-a). Results showed that the collagen concentration of 0.82 mg/mL was optimal for achieving scaffolds with high transparency and good resolution. In addition, cell viability was high after bioprinting, but cells showed rounded morphology ( Figure 5A-b). The authors argued that cells embedded into alginate bioinks were not able to degrade it and, therefore, they could not proliferate, elongate and differentiate. Thus, they proposed to add sodium citrate as a degradation system, so achieving controllable degradable scaffolds in which cell proliferation rate and epithelial specific protein expression were increased. Another study was performed by Zhang et al. [30]. They focused on simulating a corneal structure by combining two manufacturing techniques, digital light processing (DLP) and extrusion bioprinting. They applied the first technique to create an acellular supportive corneal structure using methacrylate gelatin on which different concentrations of sodium alginate/gelatin ink mixed with human epithelial cells were deposited with the extrusion bioprinter. Parameters such as geometry, thickness, mechanical properties and transparency were analyzed. Besides, cell viability was assayed. Results showed that the DLP technique significantly improved the manufacturing accuracy. The geometry, curvature and thickness of the obtained scaffold were more similar to those of the native human cornea than those obtained in the other studies carried out so far. In contrast, extrusion bio-printed layers showed high diversity among alginate/gelatin bio-ink mixtures developed in this study, in terms of printability and mechanical properties, even though, the overall transparency was good. Cell viability was high but, as the authors focused mainly on improving the manufacturing precision and geometrical control, more in vitro studies should be carried out prior to taking these scaffolds into in vivo studies. The previous studies made an approach towards manufacturing different parts of the cornea with different materials and techniques. However, to date, the only study in which the bioprinting of two different cellular parts of the cornea has been tried is that by Sorkio et al. [32]. They proposed the development of a corneal epithelium and a stroma using laser-assisted bioprinting (LaBP), and the use of stem cells, which show high differentiation capacity into epithelial stem cells and keratocytes. The authors argued that the LaBP technique may protect cells from the damage caused by nozzle stress. In addition, more viscous bio-inks can be used with this technique. In this study, two different bio-inks were developed, one based on human recombinant laminin, hyaluronic acid and human embryonic stem cells for corneal epithelium tissue, and another one composed of human collagen type I, blood plasma, thrombin and human adipose tissue derived stem cells for corneal stroma tissue. First, the scaffolds were bio-printed separately. Results showed that the bio-ink for bioprinting epithelium was printable using LaBP and that the cells embedded into the scaffolds showed high viability ( Figure 5B-b). Moreover, epithelial cell morphology and expression of corneal epithelial markers were observed after 12 days. On the other hand, although the stroma-mimicking scaffold could be fabricated through LaBP without any difficulty, scaffolds lost their shape after a few days of culture, which indicated lost mechanical strength and the necessity of a supportive structure. Cells in the stromal scaffold showed high viability and expression of proliferative markers. In addition, cell organization resembled the native human corneal stroma. On the other hand, stroma Pharmaceutics 2021, 13, 308 9 of 28 scaffolds were implanted into explanted porcine eyes, which were maintained in culture, and a proper interaction and attachment to the host tissue was observed. Finally, a scaffold containing both layers, stroma and epithelium, was bio-printed in order to simulate native human cornea. The resulting structure was opaque due to the supportive membrane that was needed to avoid the stroma layer's shape loss ( Figure 5B-a). This opacity made its application difficult as cornea substitute as it did not meet with the main corneal characteristic of being functional. Thus, other supportive structures for the bio-printed corneal scaffold were needed in order to improve both transparency and stability. which indicated lost mechanical strength and the necessity of a supportive Cells in the stromal scaffold showed high viability and expression of proliferat ers. In addition, cell organization resembled the native human corneal strom other hand, stroma scaffolds were implanted into explanted porcine eyes, w maintained in culture, and a proper interaction and attachment to the host t observed. Finally, a scaffold containing both layers, stroma and epithel bio-printed in order to simulate native human cornea. The resulting stru opaque due to the supportive membrane that was needed to avoid the strom shape loss (Figure 5B-a). This opacity made its application difficult as cornea su it did not meet with the main corneal characteristic of being functional. Th supportive structures for the bio-printed corneal scaffold were needed in ord prove both transparency and stability. Diseases affecting the corneal endothelium are the main cause of corneal transplantation [36]. Nevertheless, to date, only the study proposed by Kim et al. [46] has been focused on bioprinting. An interesting gene therapy approach was performed in which human corneal endothelial cells were genetically modified to express ribonuclease 5 (R5). R5 increases angiogenesis and facilitates endothelial cells' mitotic capacity. Therefore, cell survival rate and wound healing ability are promoted. Cells were embedded into a gelatin ink and were bio-printed by extrusion onto a bovine decellularized amniotic membrane (AM) simulating the Descement's membrane. After bioprinting, transparency, cell morphology and functional phenotype were assayed. Results showed that the scaffolds maintained their transparency for 10 days after bioprinting. Moreover, endothelial cells showed their usual shape and the R5 expression was high ( Figure 5C). Then, scaffolds were implanted into rabbit's cornea. Cell injection and an acellular membrane were used as controls. After surgery, there was an improvement in transparency in rabbits where a bio-printed scaffold had been implanted. Furthermore, four weeks later, levels of transparency near to normal cornea were achieved. In contrast, inflammation and persistent corneal oedema were observed in rabbits treated with cells and the acellular membrane. In addition, the endothelial corneal cells of the scaffold maintained their activity and shape, and rabbit native cell attachment to the scaffold was observed. Results were promising since they presented 3D bioprinting as a good alternative to the conventional treatments of corneal diseases. As we have seen, to date advances have been made in the field of 3D bioprinting to create corneal tissue, and in most of them, extrusion bioprinting is the most commonly used technique (Table 1). However, it has been demonstrated that alternative techniques, such as inkjet or laser assisted bioprinting, could also be advantageous since they can in some cases be more cell friendly. The studies are recent and are focused on the most extensive layer of the cornea, the stroma. Even so, the development of corneal epithelial and endothelial tissues also seems quite promising. Until now, due to the complex characteristics of the cornea, it has not been possible to obtain a complete multilayer corneal tissue through 3D bioprinting. Nevertheless, this technique is in its beginnings and studies have shown interesting advances over common therapies. Therefore, more research should be carried out in order to achieve a functional corneal tissue. Retina The retina is a multilayered vascularized complex tissue situated at the back of the eye, opposite to the pupil [47] (Figure 6). Its main function is to convert the light signals that reach the eye into electrical signals that are conducted to the brain [48]. This function is possible thanks to the photoreceptor cells that make up this tissue. The retina is formed by more than 130 million cells of at least 60 different types [24,49]. They are generated from the fetal retinal progenitor cells [48]. Some of these cells are rod and cone photoreceptor cells, bipolar cells, horizontal cells, retinal ganglion cells and the glial cells, among others [50]. All these cell types work together to convert the light signals into electrical signals. Under the retina there is a monolayer known as the retinal pigment epithelium (RPE) formed by pigment epithelial cells. This specialized monolayer is a physical support for the retina. In addition, it provides nutrients and growth factors that create a suitable biological microenvironment for the cells of the retina [51,52]. Retina The retina is a multilayered vascularized complex tissue situated at the back of the eye, opposite to the pupil [47] (Figure 6). Its main function is to convert the light signals that reach the eye into electrical signals that are conducted to the brain [48]. This function is possible thanks to the photoreceptor cells that make up this tissue. The retina is formed by more than 130 million cells of at least 60 different types [24,49]. They are generated from the fetal retinal progenitor cells [48]. Some of these cells are rod and cone photoreceptor cells, bipolar cells, horizontal cells, retinal ganglion cells and the glial cells, among others [50]. All these cell types work together to convert the light signals into electrical signals. Under the retina there is a monolayer known as the retinal pigment epithelium (RPE) formed by pigment epithelial cells. This specialized monolayer is a physical support for the retina. In addition, it provides nutrients and growth factors that create a suitable biological microenvironment for the cells of the retina [51,52]. The degeneration of the cells of the retina can lead to the appearance of different eye diseases. Some of these diseases are associated with a single cell type while others are associated with larger areas of the retina that involve different cell types. Some examples of the first type of disease are glaucoma [54], associated with retinal ganglion cells, and retinitis pigmentosa [55], associated with photoreceptor cells. An example of the latter is age-related macular degeneration (AMD) that arises because of mild and chronic inflammation of the central area of the retina [56]. The degenerations produced in the cells of the retina can lead to a progressive loss of vision until a total and irreversible loss is produced [27]. It is considered that many retinal diseases could be reversed if new cells from the The degeneration of the cells of the retina can lead to the appearance of different eye diseases. Some of these diseases are associated with a single cell type while others are associated with larger areas of the retina that involve different cell types. Some examples of the first type of disease are glaucoma [54], associated with retinal ganglion cells, and retinitis pigmentosa [55], associated with photoreceptor cells. An example of the latter is age-related macular degeneration (AMD) that arises because of mild and chronic inflammation of the central area of the retina [56]. The degenerations produced in the cells of the retina can lead to a progressive loss of vision until a total and irreversible loss is produced [27]. It is considered that many retinal diseases could be reversed if new cells from the retina were transplanted into the damaged area [57]. That is why many of the proposed therapies advocate the implantation of cells in the retina [58,59]. The administration of photoreceptor cells, progenitor cells, retinal sheets and RPE cells, among others, has been studied, although with not very clear results [60]. In fact, there is a considerable cell loss and a lack of control over cell behavior once implanted [61][62][63]. Therefore, appearance of abnormal behaviors and structures have been observed. As a solution to the problems involved in the implantation of cells in the retina, different scaffolds have been developed [64][65][66]. These scaffolds allow the transplantation of cells in a more controlled way. Solvent casting, electrospinning and molecular templating, among other techniques, have been used to produce these scaffolds [57]. Until now, none of the traditional microfabrication techniques have allowed the obtaining of 3D scaffolds with good structural properties and with the ability to incorporate the number of required cells in the correct position and orientation [24]. In this context, 3D bioprinting has been postulated as an excellent alternative for the design and production of scaffolds with characteristics that meet the needs of a tissue such as the retina. Some of the advantages that this technique offers are mentioned below. Cell culture. It is very difficult to seed retinal cells in vitro [67]. Many of them undergo apoptosis and those that do not undergo dedifferentiation or stop producing the signals that they produce naturally in the eye [68]. Cells used for retinal regeneration have to be properly integrated and differentiated in the case of progenitor cells, or have to remain differentiated in the case of mature retinal cells. 3D bio-printed scaffolds allow the solving of these problems since they improve the viability and the maintenance of the phenotype of the implanted cells. Moreover, these 3D scaffolds provide mechanical and physical support for the adhesion, proliferation and differentiation of the retinal cells. Complexity of scaffolds. The retina is a tissue with great structural complexity. It has several layers of diverse thicknesses and with different properties. 3D bioprinting allows the obtaining of precise scaffolds with highly complex designs (unlike previously used techniques that did not allow such control) [57]. The number of layers, their thickness and their spatial arrangement can be easily controlled by making a proper design and adjusting the printing parameters. It is necessary to replicate this spatial arrangement and thickness of the tissue so that the functions of the retina are not altered. Cell types. The retina is a tissue with a high cell diversity. In particular, it has more than 60 different cell types [24,49]. 3D bioprinting allows the incorporation of different cell types to the scaffold [69]. These cells can be incorporated into different layers, thus resembling the original tissue. In this way, a possible approach could be the incorporation of ganglion cells in a first layer, bipolar cells in a second, cones and rods in a third, and RPE cells in a fourth layer. This arrangement would be very difficult to achieve by other current manufacturing techniques. Cell orientation. The cells of the retina need a very specific orientation to be able to carry out their function properly [70,71]. The correct orientation of the cells is one of the most difficult aspects to achieve when making a scaffold. Although the new approaches that will emerge over time might allow more precise control of cell deposition, 3D bioprinting already allows a relative control over cell orientation. This can be achieved by adjusting the printing parameters, such as the printing orientation and layer thickness. Stiffness. The average stiffness of the retina is 10-20 kPa [72]. Although this parameter is not as limiting as the previous limitations, it is important when trying to create a tissue to replace the one that is damaged [73]. 3D bioprinting works with a myriad of materials and their mixtures. This allows the achieving of a stiffness as similar as possible to that of the natural retina. Taking into account all these advantages, different research groups have carried out approaches for retinal regeneration using 3D bioprinting and printing technology. Lorber et al. [74] used piezoelectric inkjet bioprinting to print retinal ganglion cell (RGC) neurons and retinal glial cells. The viability and effect of cell bioprinting on outgrowth in culture were studied. An abundant settlement of cells was detected in the nozzle. This fact significantly reduced the number of cells incorporated into the scaffolds. Nevertheless, the viabilities obtained were adequate compared to the control (69% and 78% for glial cells and 69% and 74% for retinal cells, respectively). Bioprinting did not appear to have a negative effect on the survival/regeneration properties of the cells in culture. Likewise, when used as a substrate, the glia cells that had been printed using 3D piezoelectric inkjet bioprinting retained their growth promoting properties. Kador et al. [60] used another 3D bioprinting technique to obtain their scaffold: the thermal inkjet bio-printing. In this study, they proposed a very novel approach since they evaluated the possibility of bioprinting RGC cells on an electro-spun matrix of polylactic acid. Different parameters were modified during bioprinting, such as the ejection energies and the cell densities. The results were promising. As in the previous study, good cell viabilities were obtained indicating that thermal inkjet bioprinting can effectively be used for obtaining scaffolds for retina regeneration. Furthermore, the bio-printed RGCs maintained adequate electrophysiological properties. One of the main goals of this study was to achieve a proper orientation of the cells. The microscopy images showed how the design of the matrix and the 3D bioprinting allowed the achievement of a radial arrangement of the axons of the printed cells. In this way, the orientation of the RGC was significantly improved compared to the control. Specifically, 72% of the axons were aligned with the scaffold while, in the case of the dendrites, 49% were aligned. Only 11% of the cells of the control group were aligned. Another study in which great importance was given to the orientation of the cells was that carried out by Worthington et al. [57]. In this study, different scaffolds using two photon polymerization were obtained. The studied variables were the pore size, the hatching distance, the hatching type and the slicing distance. The time necessary for printing and the fidelity of the obtained scaffolds with respect to the original designs were analyzed and optimized. In addition, induced pluripotent stem cells (iPSCs) were differentiated into retinal progenitor cells and incorporated into the printed scaffolds (Figure 7). The obtained results made it possible to clarify that those scaffolds with larger diameter pores were better for use in retinal regeneration. In these scaffolds, retinal progenitor cells could be incorporated. Besides, cells formed neural structures aligned in parallel to the vertical pores of the scaffolds. In contrast, in the scaffolds with smaller diameter pores, cells remained on top of the surface and did not align in parallel to the pores. Other studies have focused their attention on reproducing the structure of the retina. They have proposed different approaches to obtain different layers seeded with different cell types. Two examples of this type of approach are the ones proposed by Shi et al. [24,75]. Using microvalve-based bioprinting, a structure equivalent to that of the retina was created. A first monolayer was printed, made up of alginate and pluronic-containing RPE cells (ARPE-19) on a preformed membrane ( Figure 8A). This structure simulated the RPE monolayer. Over it, a second layer was bio-printed. This was also made up of the same materials and contained photoreceptor cells (Y79) ( Figure 8B,C). Two types of pattern were created for the top layer, one with a higher density of cells in the center and another with a higher density at the perimeter. The bio-ink printed with Y79 cells preserved its structure during the culture process. Viability was not compromised and cell density increased over time. This proof of concept demonstrated that a structure with characteristics similar to those of the retina can be obtained, achieving good cell viability and cytocompatibility. Wang et al. [76] made a similar approach. In this case, the 3D bioprinting technique used was laser assisted 3D. They obtained a two-layer scaffold similar to the retina. The bio-ink used in both layers was the same HA-GM (hyaluronic acid with methacrylation by glycidylhydroxyl reaction) and polyethylene-glycol-Arg-Gly-Asp-Ser peptide (PEG-RGDS). The difference between the two layers lay in the thickness and in the incorporated cells. For the RPE layer, RPE cells and a thickness of 125 um were used, while for the upper layer fetal retinal progenitor cells (fRPCs) were used that were differentiated to retina photoreceptors (PR), along with a layer thickness of 250 um. The porosity of the scaffold was analyzed as a function of the degree of methacrylation of the hyaluronic acid (low, medium or high), the swelling ratio, the rigidity of the scaffold, the viability of the cells, the formation of the two layers in the scaffold and the differentiation of fRPCs cells into PR cells within the scaffold. Results were encouraging since the stiffness of the scaffold was similar to that of the native retina, the viability remained above 70%, the microscopy images showed that two well-differentiated layers had been obtained ( Figure 8D), and the fRPCs did differentiate into PR. Therefore, it was concluded that tis co-cultivation system allowed the development of an environment similar to that of the native retina, which promoted the maturation of PRs. Pharmaceutics 2021, 13, x 16 of 28 of fRPCs cells into PR cells within the scaffold. Results were encouraging since the stiffness of the scaffold was similar to that of the native retina, the viability remained above 70%, the microscopy images showed that two well-differentiated layers had been obtained ( Figure 8D), and the fRPCs did differentiate into PR. Therefore, it was concluded that tis co-cultivation system allowed the development of an environment similar to that of the native retina, which promoted the maturation of PRs. To date, these studies have made it possible to determine: (i) the cell viability after printing; (ii) the structure of the scaffolds; (iii) the orientation of the cells within the scaffolds; and (iv) their arrangement in different layers (different levels or heights) ( Table 2). Still, much more research is needed. Among the steps to be taken in the near future are the bioprinting of more layers, the use of more cell types and the study of the ability of the bio-printed cells to transmit signals within the scaffold itself. In addition, it is expected that in the future the scaffolds obtained using 3D bioprinting will be more complex and similar to the native retina. Finally, these bio-printed tissues will serve as autologous retinal cell grafts to treat those patients suffering from retinal degeneration. To date, these studies have made it possible to determine: (i) the cell viability afte printing; (ii) the structure of the scaffolds; (iii) the orientation of the cells within the scaffolds and (iv) their arrangement in different layers (different levels or heights) ( Table 2). Still much more research is needed. Among the steps to be taken in the near future are the bioprinting of more layers, the use of more cell types and the study of the ability of the bio-printed cells to transmit signals within the scaffold itself. In addition, it is expected that in the future the scaffolds obtained using 3D bioprinting will be more complex and similar to the native retina. Finally, these bio-printed tissues will serve as autologous retinal cell grafts to treat those patients suffering from retinal degeneration. Conjunctiva The conjunctiva is a mucosal tissue that covers the sclera and provides lubrication and protection to the eye by producing tears and mucus [77]. It consists of a goblet cell rich in highly vascularized stratified epithelium [78]. This tissue suffers from different injuries caused by ocular thermal or chemical abrasions, conjunctival lacerations, autoimmune diseases, inflammation, foreign bodies or surgery, etc. [79]. To treat the damaged conjunctiva the usually employed strategies are surgery and autologous grafts or allograft tissues such as AM and pericardium [80]. These approaches have some limitations: (i) unavailability of healthy conjunctiva (in the case of autologous grafts); (ii) immune responses; (iii) keratinization; (iv) goblet cell loss; (v) microbial infections; (vi) low level of stratifications, and (vii) opacification of the site, etc. [78]. These limitations make necessary the development of new strategies among which 3D printing has been postulated as an excellent alternative. 3D printing can act as an improvement for the development of structures similar to the conjunctiva as it allows the achievement of the following characteristics. Thickness. It is crucial to develop a 3D structure with an adequate thickness; that is, as close as possible to the original (average thickness of 33 µm). By determining the number of layers and their thickness, the size of the 3D printed structure can be controlled very precisely. Cell density. 3D bioprinting allows control of the density of goblet cells included in the 3D developed membrane. Transparency, elasticity and slight rigidity. The large number of materials that can be used in 3D bioprinting allows the controlling and adjusting of the color and transparency of the construct. These materials also make possible the obtaining of elastic properties similar to that of the healthy tissue. In addition, the rigidity of the scaffold can also be adjusted. One of the problems related with the grafts used so far is the difficulty of handling in operations because of their fineness. 3D bioprinting makes it possible to achieve scaffolds with sufficient rigidity to facilitate manipulation, without this, in turn, having a negative biological effect. Biological activity. The use of materials of biological origin allows the adjustment of the re-epithelization capacity, decreasing scar formation and fornix foreshortening, adjusting biodegradability and achieving good biocompatibility. So far, the only approach that has used 3D printing to obtain a structure that allows regeneration of the damaged conjunctiva is the one carried out by Dehghani et al. [81]. They developed a membrane by extrusion 3D printing using gelatin, elastin and hyaluronic acid as materials. They carried out multiple rheological and texturometry tests to guarantee that the ink and the membrane obtained had the necessary properties to be implanted in the conjunctiva. Biological properties were also analyzed in terms of cytocompatibility, adhesiveness and cell proliferation in vitro, and epithelialization, inflammation, scar tissue formation and presence of granulation tissue in vivo. All these results were compared with an AM (frequently used in injuries of conjunctiva). The results obtained were promising. The ink could be properly printed and the obtained membrane had adequate color and transparency, and its handling was simpler than that of the AM. Furthermore, in vitro good cytocompatibility, adhesion and cell proliferation were obtained. In vivo, the results were also positive. The epithelization time was similar in the printed membrane and in the AM. However, the data regarding inflammation, cell density, degradation and granulomatous reaction were better in the 3D printed membrane group. These results show that this technique is suitable for the development of membranes for the regeneration of damaged conjunctiva. In this sense, it is reasonable to think that in the near future more research into the application of 3D printing to conjunctiva regeneration will be carried out. 3D Printing for Ocular Drug Delivery The main objective in the field of pharmacology is to achieve the maximum therapeutic effect with the minimum toxicity [82]. In this regard, the development of personalized Pharmaceutics 2021, 13, 308 21 of 28 patient specific drugs or doses is booming. 3D printing has recently been postulated as the appropriate technology to accomplish this goal due to its ease of use, fast speed, and accessibility [83]. In this way, the possibility of developing pharmaceutical forms containing various drugs, adjusting the doses to each patient and modifying drugs' pharmacokinetic profiles has become a reality in preclinical studies thanks to 3D printing technology [83]. This personalized medicine would bring a huge advantage to patients suffering from chronic eye pathologies, since current treatments require the constant application of eye drops and, in the worst cases, repetitive intravitreal injections that can cause devastating intraocular inflammation [84]. In this context, although there have been several studies in which 3D printing devices have been developed as drug delivery systems, only the one proposed by Won et al. [84] focuses on the eye. In this study, a flexible coaxial printing was used in order to develop a system that contained two drugs for the treatment of retinal vascular disease (RVD), which is based on an abnormal vascularization of the retina. The drug delivery system consisted of an external shell composed of polycaprolactone and bevacizumab (PCL-BEV), a drug that prevents excessive angiogenesis. The interior part was composed of sodium alginate and dexamethasone (ALG-DEX), an anti-inflammatory drug. In vitro assays showed a continuous release of BEV and DEX. Moreover, good biocompatibility and a reduction in the growth rate of human umbilical vein endothelial cells was observed. Then, the printed drug delivery system was intravitreally injected into two animal models: rabbits, in order to study the release kinetics, and choroid neovascularization (CVC) rat models, to determine its therapeutic effects. Intravitreal drug injections were used as control. Results corroborated that the printed drug delivery system prolonged the release of BEV and DEX compared to control. Interestingly, higher angiogenic inhibition over time was observed in CVC rats compared to controls and a reduction of inflammation was achieved. As reflected in this study, the implementation of 3D printing technology for the development of specific drug delivery systems in ophthalmology can be valuable. In addition, the manufacture of artificial tissues similar to native tissues with this technology can be useful when it comes to screening new drugs for eye diseases. However, before this technology can be extensively applied in clinics, multiple regulatory questions should be addressed. Ethical Issues and Commercialization Regulatory Aspects Although the previously mentioned technical requirements can be fulfilled by increased research knowledge, there are still some relevant concerns related to both ethical and regulatory aspects that jeopardize the road to clinic of this "game changer" technology [1,2]. From an ethical point of view, a clear benefit of this technology includes the use of 3D bio-printed ocular organs or tissues for academic or research applications, as intermediate drug testing models between in vitro conditions and in vivo probing of concepts. Such models could work as a promising alternative to minimize the use of animals in the laboratory. In addition, patients' own cells reprogrammed into induced pluripotent stem cells (iPSCs) and bio-printed into 3D ocular structures could represent a more efficient and reliable model for drug testing than the use of experimental animals. However, there are also risks and ethical issues that should be considered, especially when 3D bioprinting technology is aimed at tissue engineering purposes. In this case, the composition of biological inks raises some concerns, not only associated with the security of the grafts implanted, but also related to the biological origin of such cells [3]. In the clearest scenario, where patient's autologous cells are included in the bio-ink composition, a random migration of cells from implanted ocular grafts could arise in different parts of the body, leading to potential undesired effects [5,8]. It is also likely that the biological behavior of such cells can be altered due to the mechanical stress that cells suffer by the transient forces applied during the bioprinting process. Therefore, although 3D bioprinting technology holds huge potential in tissue engineering clinical practice, the benefit-risk balance should be considered, in the same way as in other advanced therapies such as cell and gene therapy. In case non-autologous cells are implanted, apart from previously mentioned concerns, the possibility of inflammatory response against implanted ocular graft also needs to be analyzed. In this scenario, it should not be forgotten that the donor of cells for the implementation of 3D bio-printed grafts needs to sign an informed consent to allow the use of such cells. The situation is more delicate if non-autologous stem cells at the embryonic status are manipulated and bio-printed to become part of a grafted ocular tissue or organ, due to the ethical dilemma that arises with the use of embryonic cells. Another more challenging possibility is the use of cells derived from animal models to print ocular tissues implanted in humans. In this case, in addition to the previously mentioned biological concerns, the risk of developing zoonosis diseases should also be considered [20]. However, not only the origin and status of bio-printed cells rise ethical issues. We should also consider if any biological "item" can be printed [21]. For instance, and leaving apart any technological consideration, strictly from an ethical point of view, when implanting bio-printed tissues, a retina can be more problematic than a cornea, due to the neuronal structure of the retina and its direct connection to specific areas of the brain. This issue also generates the discussion as to whether specific areas of the brain, or even the brain or the eye as a whole, could be bio-printed and implanted into human beings. Another ethical concern for the application of 3D bioprinting technology into regular clinical practice arises from the design and implementation of clinical trials for personalized medicines. Traditionally, clinical trials are designed and classified into different phases to evaluate the safety of the treatment in the early stages, and later the efficacy in a large population, before the introduction of a drug into the market. However, this approach is not feasible in the case of patient-tailored medical products for ophthalmic purposes, due to interindividual variability among human beings, which hampers any extrapolation of the results obtained in a specific patient, although experience accumulated in clinical cases could serve to gain progressive knowledge in order to apply the technology in medical practice [22]. Furthermore, the irreversible nature of grafts implanted by bioprinting technology impedes the patient's withdrawal from the trial after implantation, in the case of complications. In this sense, the development of 4D bio-printable organs, that can be biodegraded under physiological conditions after having performed the desired effect, merits special attention [24,25]. However, it should be also borne in mind that enrolling in this kind of clinical trial involves the acceptance of high risks associated with the implementation of the technology. Therefore, only advanced stages of diseases that have been unsuccessfully treated with conventional approaches should be amenable for this kind of treatment. When dealing with health and financial resources, it is logical to expect ethical conflicts. In this sense, the technological possibility to bio-print and implant specific organs or tissues as an alternative solution to face advanced stages of diseases could stratify society. In this sense, even though the manufacturing of bio-inks and 3D bio-printing are not expensive items, the multidisciplinary nature of the global process, as well as the surveillance of this approach, makes it a high-cost procedure [85]. Hence, its implementation would be not affordable for all strata of society, only being obtainable for a particular subgroup of the total population with access to financial resources. Therefore, only those who can afford to pay for the bio-impression of their own organs would presumably enjoy a longer and better quality of life, minimizing for instance the use of immune-suppressant drugs indicated to avoid rejections of conventionally transplanted organs. In addition, relevant organs or tissues specifically designed with better biological properties could be implanted not only to deal with advanced stages of diseases, but also to enhance their physiological performance, which at the end could result in the elaboration of "super" bio-items for eugenic goals [86], such as "super" eyes or retinas. Considering all the risks and benefits that 3D bioprinting technology can offer in the coming years, not only to the research and medical community but also to patients affected by ocular diseases, legal and regulatory aspects related to the implementation of the technology need also to be also deeply analyzed to avoid illicit and fraudulent use if the technology finally ends up in the "wrong hands". However, actually, the scaleup of 3D bioprinting technology and its clinical application for medical purposes does not fit into any of the current regulatory categories, despite that, from a global point of view, it could be considered as a specific tissue engineering approach [87]. In addition, the patient-tailored application of the technology hampers the compliance with global regulatory requirements for commercialization purposes that at moment are limited to recommendations, notifications and reports provided by European and American agencies, the EMA and FDA, respectively. Although nowadays both agencies aim to legislate the application of this innovate technology into clinical practice, they lack a specific regulatory framework [88]. From a more world-wide point of view, only a few regulatory agencies in countries such as Japan and South Korea have developed broad regulatory measures that can be applied to 3D bioprinting technology. This initiative could work as a starting point and reference for other countries. In any case, such regulatory measures are mainly focused on the application of these technologies for academic and research purposes or for the development of acellular devices in ophthalmology such us spectacles, lenses or smartphone-based fundus cameras, etc. [88]. Therefore, in order to implement this technology regularly in medical practice, it is necessary to start working on the design of a multidisciplinary and world-wide panel to provide a global regulatory framework. It also should be borne in mind that, in order to find a solution and resolve legal problems associated with the technology, all related tenets need to be addressed. In this sense, due to the complexity of the technology, a "whole" legal approach would be preferred, rather than a "piecemeal" approach. Current Challenges and Future Perspectives of Ocular 3D Bioprinting Recent advances in ocular 3D bioprinting have brought new opportunities for eye tissue engineering with potential biomedical applications, and nowadays it seems unquestionable that this field will continue to grow and evolve in future years. However, there are still relevant challenges to overcome before ocular and, in general, 3D bioprinting becomes a real clinical option. First, the materials for bio-ink preparation should be biologically functional while maintaining robust and controllable post-printing mechanical properties and should enable adequate physiological, bio-chemical and mechanical interactions with the cellular component [89]. One strategy to find a desirable balance between biological activity and mechanical characteristics is the use of hybrid constructs that contain, on the one hand, synthetic materials that provide structural integrity and, on the other hand, natural materials that provide a cell-friendly environment for cellular growth. Other approaches are based on the chemical modification of the scaffold or on the use of synthetic peptides such as Arg-Gly-Asp (RGD) in order to prompt the crosslinking of the material and control its mechanical properties or its degradation time. For instance, a similar strategy was used for retinal bioprinting using a bio-ink based on HA with methacrylation by glycidyl-hydroxyl reaction and PEG-RGDS, with encouraging results, since the mechanical properties achieved allowed the bioprinting of two retinal layers with suitable rigidity and high cell viability [76]. In addition, cell differentiation occurred within the scaffold, which indicated that the hydrogel was biologically active and enabled biochemical interactions with the cell component. In this regard, appropriate cell orientation within the scaffold has also been achieved in retinal bioprinting using retinal ganglion cells that maintained their functional electrophysiological properties after being printed, demonstrating that the scaffold provided suitable physiological, biochemical and mechanical interactions for that purpose [60]. Cell sourcing constitutes another important challenge for 3D bioprinting of tissues and organs. In fact, a high number of regeneration-competent cells are needed and, due to tissue heterogeneity, different types of cells are also required. In fact, most organs are more complex than current 3D bioprinters can reproduce and achieving their intricate composition and functionality is still far from clinical translation [4,90]. In eye 3D bioprint-ing, most ocular structures bio-printed so far are limited to one or two cell types and, in the case of cornea and retina, only one or two layers have been bio-printed in the same construct, which is far from the authentic complex configuration of these ocular structures. In any case, regarding the challenges related to limited cell availability in forming the scaffolds, some of the solutions carried out to date are based on the use of stem cells or progenitor cells such as mesenchymal cells or induced pluripotent stem cells (iPSC), which, ideally, should be autologous or non-immunogenic [89]. In this sense, human fetal retinal progenitor cells have been used for retinal bioprinting and successful differentiation into photoreceptor cells was achieved [76]. In addition, human iPSC have also been used for retinal bioprinting and differentiated into retinal progenitor cells [57]. Considering all this, important advances have been made in order to overcome specific challenges related to cell sourcing, but more fundamental research is still needed in order to be able to bio-print a whole functional ocular structure such as the cornea or the retina with all the required different cell types and layers and the necessary biomechanical cues. These advances would contribute in the future to mimicking the complexity of the desired organ structures in a more precise manner. In addition to the difficulties in obtaining a highly complex and similar to native bio-printed construct, another key aspect to consider before clinical translation could occur would be the vascularization and innervation of such tissue-engineered constructs after transplantation. So far, most tissue constructs obtained by 3D bioprinting lack a functional vascularization network, which would hinder the oxygen and nutrient supply after implantation in vivo [89]. Different strategies have been postulated in order to solve this issue in different types of tissue, including the use of angiogenic growth factors, the embedding of microchannels that would enhance the diffusion of oxygen and nutrients, or the direct fabrication of vasculature [89]. However, the reproduction of the complex and complete vascular network necessary for clinical translation remains very challenging. Regarding the eye, bioprinting a full ocular structure has still not been accomplished, so that would be the first step before moving to vascularization and innervation of the tissue construct. However, this aspect would acquire special relevance in the case of the retina, where the light signals captured by the photoreceptors must be processed into electrical impulses and transmitted through the optic nerve to the brain. In this regard, the functional electrophysiological properties of bio-printed retinal ganglion cells achieved so far [60] hold promise for future functional retinal constructs with appropriate connections among cells of the different retinal layers, capable of completing the complex visual phototransduction process. In summary, recent advances in the field of 3D bioprinting and, particularly, for ocular tissue-engineering show promise for future biomedical applications, although there are still many challenges that need to be overcome before clinical translation can occur. Mimicking the complexity and heterogeneity of organs and providing them with the diversity of functional and supporting cell types, as well as with the essential functional elements such as vasculature and innervation, represent the major difficulties that 3D bioprinting is currently facing. In addition, for future commercialization, further aspects such as standardization of protocols, regulation of the process, the cost-effectiveness for scaling and the logistics of 3D bio-printed products would need to be taken into consideration. The key benefits of 3D bioprinting include the possibility of more targeted and personalized medicine, automated tissue fabrication and the flexibility of incorporating a wide variety of cells and materials in a precise anatomical 3D geometry. Further research for in vitro optimization and in vivo implementation of bio-printed tissue constructs and the combined efforts of different multidisciplinary fields would enhance the progress of 3D bioprinting towards clinically relevant bio-printed organs. Conclusions In recent years, great advances have been made while using 3D bioprinting for eye tissue engineering. Different ocular tissues with various layers and various cell types have been replicated. Nevertheless, we are still far from achieving complete tissues similar to healthy ones. It is expected that, in the future, the different ocular components will be completely replicated using 3D bioprinting. This progress could allow the combination of the structures obtained and to integrate them into a single construct, as a complete ocular model [27], which would require the consideration of corresponding ethical and biological aspects. Furthermore, as previously mentioned, the generated models would allow planning of operations before performing them, understanding the interaction between cells and the progression of different diseases affecting the tissue, or analyzing the effects of diverse drugs.	2021-03-04T06:16:47.330Z	2021-02-26T00:00:00.000	{ "year": 2021, "sha1": "36a0abbd3038c8ecaadf236a1e4b40bde35db531", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1999-4923/13/3/308/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "e17a1a309c4f0f9af8020c30e187fb70dbfbf4c6", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
14842392	pes2o/s2orc	v3-fos-license	Number Fluctuation and the Fundamental Theorem of Arithmetic We consider N bosons occupying a discrete set of single-particle quantum states in an isolated trap. Usually, for a given excitation energy, there are many combinations of exciting different number of particles from the ground state, resulting in a fluctuation of the ground state population. As a counter example, we take the quantum spectrum to be logarithms of the prime number sequence, and using the fundamental theorem of arithmetic, find that the ground state fluctuation vanishes exactly for all excitations. The use of the standard canonical or grand canonical ensembles, on the other hand, gives substantial number fluctuation for the ground state. This difference between the microcanonical and canonical results cannot be accounted for within the framework of equilibrium statistical mechanics. After the experimental discovery of BEC in a trapped dilute gas at ultra-low temperatures, much attention has been paid to the problem of number fluctuation in the ground state of the ideal system [1,2,3,4,5,6,7,8,9], as well as a weakly interacting bose gas [10,11,12,13,14,15,16,17]. There are also a few papers on ground state fluctuations in a trapped fermi gas [18]. There are several reasons for this interest. In standard statistical mechanics, number fluctuation is related to density-density correlation, and to the compressibility of the system in the grand canonical ensemble (GCE) [17]. Light scattering cross section off the medium, in principle, may be related to the ground state fluctuation in the system [19]. The so called grand canonical catastrophe in an ideal bose gas, where the fluctuation diverges at low temperatures, was already known [20]. Therefore a more accurate treatment of the problem was needed for trapped gases. In the microcanonical treatment of number fluctuation from the ground state in a harmonic trap, the problem is closely related to the combinatorics of partitioning an integer, and thus there was an interesting link to number theory [1]. It turned out that the result for the ground state number fluctuation was very sensitive to the kind of asymptotic approximations that are made. Another aspect that drew much attention in the literature was the difference in the calculated results for fluctuation using the canonical and the microcanonical formulations [7,8]. It was pointed out by Navez et al. [7] that for a trapped bose gas below the critical temperature, the microcanonical result for fluctuation could be obtained solely using the canonically calculated quantities, which in turn may be obtained from the so called Maxwell Demon ensemble [21], to be explained later. In this paper, we give an example of a quantum spectrum that has no number fluctuation in the ground state for any excitation energy in the microcanonical ensemble, as a direct corollary of the fundamental theorem of arithmetic. The canonical ensemble, on the other hand, yields a dramatically different ground state number fluctuation. The method of Navez et al. fails to account for the difference between the microcanonical and canonical results in this example. This failure of the canonical (grand canonical) ensembles is due to the peculiar nature of the single-particle spectrum in our example. Generally, when a large excitation energy is supplied to a system, there are a very large number of distinct microscopic configurations that are accessible to it. All these different microstates describe the same macro-state of a given excitation energy. The classic example is that of bosons in a harmonic trap, where the number of partitions of an integer number, corresponding to the number of microstates, increases exponentially. We use, on the other hand, another example from number theory, to propose a system where the excitation energy, no matter how large, is locked in one microstate. Consequently, although it is possible to explicitly calculate the canonical or grand canonical partition functions and therefore the thermodynamic entropy for this example, it does not approach or equal the information theoretical entropy that can be exactly calculated using number theory [22]. So far as we know, this constitutes a novel and new example that links number theory and thermodynamics, and gives radically different results for the microcanonical and canonical ensembles. We consider bosons in a hypothetical trap with a single-particle spectrum (not including the ground state, which is at zero energy) where p runs over the prime numbers 2, 3, 5, ... Of course, such a spectrum is not realizable experimentally, and it is merely a means of performing a thought experiment. We shall use, in what follows, both a truncated sequence of primes, as well as the infinite sequence when we perform the canonical calculations for fluctuation. First, however, we perform the exact calculation for number fluctuation from the ground state. Suppose that there are N bosons in the ground state at zero energy, and an excitation energy E x is given to the system. In how many ways can this energy be shared amongst the bosons by this spectrum ? Before giving the answer, we remind the reader of the fundamental theorem of arithmetic, which states that every positive integer n can be written in only one way as a product of prime numbers [23] : where p r 's are distinct prime numbers, and n r 's are positive integers including zero, and need not be distinct. It immediately follows from Eq. (2) that if the excitation energy E x = ln n, where the integer n ≥ 2, there is only one unique way of exciting the particles from the ground state. If E x = ln n, the energy is not absorbed by the quantum system. Since the number of bosons excited from the ground state, for a given E x , is unique for this system, the number fluctuation in the ground state is identically zero ! Moreover, this conclusion is valid whether we take in Eq. (1) an upper cut-off in the prime number, or the infinite sequence of all the primes. The information-theoretic entropy at an excitation energy E x is where P i is the probability of excitation of the microstate i. Since only one microstate contributes with unit probability, and all others have P i = 0 (when E x = ln n), the entropy S = 0. It is also straight forward to calculate the ground state population N 0 as a function of the excitation energy E x . For this purpose, we truncate the spectrum given by Eq. (1) to the first million primes, with a cutoff denoted by p ⋆ , and take N = 100. We shall display the numerical results after describing the canonical calculations. Our next task is to calculate these quantities in the canonical ensemble, and see if the differences in the microcanonical and canonical results may be accounted for (as N → ∞) using the method of Navez et al. [7]. We first do the calculation for the truncated spectrum. The one-body canonical partition function is then given by Z 1 (β) = 1 + p * p=2 exp(−β ln p). The N-body bosonic canonical partition function is obtained by using the recursion relation [24] Once Z N is found, the ground state occupation N 0 = n 0 N and the ground state number fluctuation for the canonical ensemble can be readily be computed [18,25]. We define δN 2 0 = ( n 2 0 N − n 0 2 N ), where the RHS is calculated using Eqs. (24) and (25) of Ref. ([18]). In Fig. 1a) and b), we display the results of the canonical calculations for the ground state occupancy fraction N 0 /N and the ground state fluctuation (δN 2 0 ) 1/2 /N for N = 100 as a function of temperature T with the truncated spectrum of the first million primes. For comparison, we also show the results of the corresponding grand canonical calculation. The grand canonical catastrophe for the number fluctuation is clearly evident. It is also easy to calculate the canonical (equilibrium) entropy S = ln Z N (β) + βE x . The comparison with the combinatorial (or microcanonical) results requires that we convert the canonical temperature to excitation energy E x . This is done by using the standard relation E x = − ∂ ln ZN (β) ∂β . In Fig. 2, the canonical fractional occupancy of the bosons in the excited states, N e /N , for N = 100, is compared with the (exact) combinatorial (or microcanonical) calculation as a function of the excitation energy E x . Although the canonical and the grandcanonical N e /N are nearly identical, the corresponding microcanonical quantity is radically different. This anomaly persists even as N → ∞, showing the breakdown of the equivalence between the microcanonical and the other ensembles. In Fig. 3, we display the behaviour of the canonical entropy as a function of T and E x . The microcanonical entropy S(E x ), of course, is zero, and so is the number fluctuation δN 2 0 . Thus we find that the canonical results have no resemblance with the exact microcanonical ones. All these calculations were performed for a truncated spectrum (first million primes) of ln p, as specified earlier, and for N = 100. We shall now use the procedure of Navez et al. [7] to check if the microcanonical results may be obtained from a canonical calculation as N → ∞. These authors constructed the so called Maxwell's Demon ensemble in which the ground state (for T < T c ) was taken to be the reservoir of bosons that could exchange particles with the rest of the subsystem (of the excited spectrum) without exchanging energy. Denoting the grand canonical partition function of the excited subsystem by Ξ e (α, β), with α = βµ, it was shown that the canonical occupancy of the excited states, N e , and the number fluctuation (δ 2 N e ) could be obtained from the first and the second derivative of Ξ e with respect to α, and then putting α = 0. It was further noted that the microcanonical number fluctuation for the excited particles was related to the canonical quantities by the relation where the superscript ∞ denotes N → ∞. This worked beautifully for harmonic traps in various dimensions. These calculations, for our system, are also easily done for β > 1. We now consider the spectrum (1) to be the infinite sequence of the primes, and evaluate the RHS of Eq. (6). We readily obtain the convergent expression (for β > 1) The RHS of Eq. (6) is nonzero, and therefore does not agree with the microcanonical result. The failure of the above formalism of Navez et al. [7] is because of the very special nature of the single-particle spectrum (1). Consider constructing the N-particle canonical partition function Z N (β) from this spectrum, as we had done. As N → ∞, a little thought will show that Z N → ζ(β), where ζ(β) = n 1 n β is the Riemann zeta function. This is because we are allowed to span over all E in calculating Z N (β). Similarly, the grand partition function Ξ e (α, β) with α = 0 is none other than the Euler product representation of the Riemann zeta function [23]. This has a density of states growing exponentially with E, and has been studied in connection with the limiting hadronic temperature [26]. By contrast, the number of accessible states with a fixed N and E in the isolated microcanonical set-up does not increase at all. Since the energy remains locked in one microstate, the isolated system cannot be described through the usual concepts of statistical mechanics. This research is supported by NERSC (Canada). The authors would like to thank David M. Cooke and Jamal Sakhr for helpful discussions. 2: a) Plot of the average occupancy in the excited states Nex/N , for N = 100, versus the excitation energy Ex in the canonical ensemble (continuous bold curve), compared with the exact microcanonical calculation. The latter calculation is done for Ex = ln n, where n is an integer, and the results are shown by dark points. These are joined by dotted lines to emphasize their zigzag character. For example, the sixth point (including 0) corresponds to Ex = ln 6, and gives Ne = 2, corresponding to the prime factor decomposition 2 × 3.	2017-04-08T17:02:33.017Z	2002-10-28T00:00:00.000	{ "year": 2002, "sha1": "bd0c92f6ac44adb64b58b5c59aae1cbcb9ce1a0f", "oa_license": null, "oa_url": "http://arxiv.org/pdf/cond-mat/0210624", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "7721058b165bc8ef67f87aaeb9c077f1604dce21", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Mathematics", "Medicine", "Physics" ] }
5051136	pes2o/s2orc	v3-fos-license	Mechanisms Underlying the Risk to Develop Drug Addiction, Insights From Studies in Drosophila melanogaster The ability to adapt to environmental changes is an essential feature of biological systems, achieved in animals by a coordinated crosstalk between neuronal and hormonal programs that allow rapid and integrated organismal responses. Reward systems play a key role in mediating this adaptation by reinforcing behaviors that enhance immediate survival, such as eating or drinking, or those that ensure long-term survival, such as sexual behavior or caring for offspring. Drugs of abuse co-opt neuronal and molecular pathways that mediate natural rewards, which under certain circumstances can lead to addiction. Many factors can contribute to the transition from drug use to drug addiction, highlighting the need to discover mechanisms underlying the progression from initial drug use to drug addiction. Since similar responses to natural and drug rewards are present in very different animals, it is likely that the central systems that process reward stimuli originated early in evolution, and that common ancient biological principles and genes are involved in these processes. Thus, the neurobiology of natural and drug rewards can be studied using simpler model organisms that have their systems stripped of some of the immense complexity that exists in mammalian brains. In this paper we review studies in Drosophila melanogaster that model different aspects of natural and drug rewards, with an emphasis on how motivational states shape the value of the rewarding experience, as an entry point to understanding the mechanisms that contribute to the vulnerability of drug addiction. INTRODUCTION From insects to humans, organisms living in complex environments need to respond quickly and appropriately to different stimuli by choosing one action over another to increase their chances of survival and reproduction. Reward systems play a key role in promoting this aim by motivating animals to repeat behaviors that increase their fitness, such as eating, drinking, sexual interaction, and parental behaviors. Drugs of abuse affect the same brain regions used for the processing of natural rewards, creating the pleasurable feeling indicative of a fitness benefit, and with repeated use can lead to compulsive drug abuse and addiction (Nesse and Berridge, 1997;Koob, 2009). The American Psychiatric Association defines addiction as "maladaptive pattern of substance use manifested by recurrent and significant adverse consequences related to the repeated use of substances" (American Psychatric Association, 2013). This is characterized by a sequence of stages: (1) initial voluntary consumption of the drug, accompanied by an acute hedonic drug response, (2) repeated use, leading to compulsive and uncontrolled drug use, and finally, (3) physical and mental dependence (Koob and Bloom, 1988;Wolffgramm and Heyne, 1995;Koob, 1997Koob, , 2009Nesse and Berridge, 1997). Understanding the complex nature of human addiction is one of the greatest challenges in contemporary neuroscience, requiring parallel efforts of many scientific disciplines. One important approach is the use of animal systems to model certain features of the process, such as the reinforcing properties of drug rewards. Early studies by Karl von Frisch demonstrated the ability of sugar reward to reinforce preference for certain colors in honey bees (von Frisch, 1914). Subsequent studies by Olds and Milner demonstrated that rodents can learn to press a lever to receive intracranial self-stimulation (ICSS), facilitating the discovery of brain areas that encode reward (Olds and Milner, 1954). These seminal studies paved the path for the development of complex behavioral paradigms that measure the rewarding effects of drugs. Examples include self-administration paradigms, in which voluntary lever pressing results in delivery of a drug dose (Weeks, 1962;Thompson and Schuster, 1964), and conditioned place preference, where animals learn to associate a certain environment with receiving a drug, and the preference for this environment is tested afterwards in the absence of the drug (Rossi and Reid, 1976). Although the existing models do not entirely recapitulate the complexity of human addiction, they model important features of drug addiction (Koob, 2009;Lynch et al., 2010). For example, the positive reinforcing actions of binge intoxication is captured using self-administration paradigms in rodents and monkeys (Johanson and Balster, 1978;Collins et al., 1984), while the negative reinforcing properties of the withdrawal phase are measured by increased anxiety-like responses (Sanchis-Segura and Spanagel, 2006). The craving stage can be modeled by "cue-induced reinstatement, " in which the reinstatement of drug seeking is tested after the induction of drug cues following drug self-administration training (Sanchis-Segura and Spanagel, 2006;Liu et al., 2008;Mantsch et al., 2016). Although it is more common to use mammals to study addiction, insect behavior is no less organized and driven by reward. Many studies over the years have established the fruit fly Drosophila melanogaster as a non-conventional but very relevant model to explore molecular mechanisms underlying drug response. These have mostly focused on ethanol, modeling early stages of ethanol exposure, including its immediate locomotor effects (reviewed extensively in Rodan and Rothenfluh, 2010;Kaun et al., 2012;Ghezzi et al., 2013a), its hedonic value, as reflected by voluntary consumption (Devineni and Heberlein, 2009), and the formation of longlasting memories for the rewarding experience (Kaun et al., 2011; Figure 1). This review will present recent progress in which fruit flies were used to uncover genetic and environmental elements that influence the likelihood of progressing from initial exposure to repeated drug use. It will focus on drug-oriented studies and those that are not drug oriented but share mutual mechanisms and principles with addiction, such as learning and memory, and neuronal mechanisms that encode and process natural rewards. Together, the cellular pathways, neuronal circuits and newly discovered principles that govern reward processing can serve as a conceptual framework for understanding the mechanisms that underlie the risk to develop addiction. METHODS OF STUDYING ETHANOL RELATED BEHAVIORS IN FRUIT FLIES Flies encounter ethanol in their natural habitat, and as such, acquired many adaptations that enable them to survive and thrive in ethanol-rich environments (Gibson et al., 1981). Flies exhibit natural preference for ethanol: the smell of ethanol was shown to be an attractive cue using olfactory trap (Reed, 1938;Dudley, 2002;Devineni and Heberlein, 2009), and females show preference to lay eggs on ethanol containing substrates (Siegal and Hartl, 1999;Azanchi et al., 2013;Kacsoh et al., 2013). Flies develop preference to consume ethanol-containing food in a two-choice consumption paradigm. The kinetics of their preference and its extent depend on genetic background (Merçot et al., 1994;Devineni and Heberlein, 2009), prior exposure to ethanol (Peru y Colón de Portugal et al., 2014), sampling time (Devineni and Heberlein, 2009), and prior sexual experience (Shohat-Ophir et al., 2012). Importantly, flies display similar behavioral responses to acute exposure to ethanol as mammals: increased motor response when exposed to a low dose of ethanol, and sedation when reaching higher doses . Repeated exposure to ethanol results in functional tolerance and increases the time and dose needed to induce sedation. This reflects neuronal plasticity that corresponds to tolerance (Figure 1), but can also be caused by changes in ethanol metabolism (Scholz et al., 2000). Over the years there have been several experimental systems to study the locomotor response to ethanol intoxication, the first of which was the inebriometer system (Cohan and Graf, 1985;Cohan and Hoffmann, 1986;Weber, 1988), in which flies lose their postural control when exposed to ethanol vapor. The system was later adapted for high throughput functional genetic screens by the Heberelin lab (Moore et al., 1998), and was subsequently replaced by video tracking systems that measure changes in fly velocity during acute intoxication, and assays that measure loss of righting response when reaching sedating levels (Wolf et al., 2002;Maples and Rothenfluh, 2011). Many genes and cellular pathways in neurons and glia cells have been shown to modulate the sensitivity of flies to both the positive and negative motor responses upon exposure to ethanol vapor, and the development of tolerance (Moore et al., 1998;Scholz et al., 2000Scholz et al., , 2005Berger et al., 2004;Ghezzi et al., 2004Ghezzi et al., , 2013bCorl et al., 2005;Cowmeadow et al., 2006;Kong et al., 2010;King et al., 2011King et al., , 2014Kapfhamer et al., 2012;Krishnan et al., 2012Krishnan et al., , 2016Li et al., 2013;McClure and Heberlein, 2013;Pohl et al., 2013;Troutwine et al., 2016). Some of the identified fly genes, pathways and principles paved the way for parallel studies Oviposition preference for a substrate containing ethanol (Azanchi et al. 2013) Acute ethanol sensitivity, loss of postural control: Inebriometer (Moore et al.1998), Locomotor tracking assay monitoring hyperactivity and sedation (Wolf et al. 2002) Ethanol tolerance upon repeated exposure (Scholtz et al. 2002) FIGURE 1 \| Schematic illustration of the behavioral paradigms that are used to model different features of drug addiction in Drosophila. in mammals (Corl et al., 2009;Lasek et al., 2011a,b,c;Maiya et al., 2012Maiya et al., , 2015Kapfhamer et al., 2013). A breakthrough in modeling aspects of drug reward in flies was the introduction of two paradigms: a conditioned response to ethanol vapor (Kaun et al., 2011), and a two-choice assay to measure voluntary ethanol consumption (Ja et al., 2007;Devineni and Heberlein, 2009). In the first paradigm, flies learn to associate cues with ethanol intoxication and develop longlasting attraction for an ethanol-paired cue (Kaun et al., 2011). A demonstration for the relevance of this model as a system to study aspects of drug reward was the finding that flies are willing to tolerate electric shock in order to approach an odor cue predicting ethanol reward (Kaun et al., 2011). The two-choice ethanol consumption paradigm measures motivation to obtain drug rewards, where flies can choose to feed from ethanol or non-ethanol containing food in a capillary feeder system (CAFE) (Devineni and Heberlein, 2009;Pohl et al., 2012;Shohat-Ophir et al., 2012;Xu et al., 2012;Ojelade et al., 2015;Zer et al., 2016). Another two-choice ethanol consumption paradigm is the FRAPPE (Fluorometric Reading Assay of Preference Primed by Ethanol), a novel assay based on the CAFE system, which allows precise and high throughput measurement of consumption in individual flies (Peru y Colón de Portugal et al., 2014; Figure 1). Lastly, a recent study by Shao, et al. established a new reward selfadministration paradigm that is based on optogenetic stimulation of neurons that encode positive valence (Shao et al., 2017). In this assay, flies harboring the red shifted channel rhodopsin CsChrimson (Inagaki et al., 2014a) in NPF neurons prefer to be in a zone that triggers optogenetic stimulation of their NPF expressing neurons (Shao et al., 2017; Figure 1). Although this assay does not measure drug related responses, it facilitates the identification of neurons that induce immediate pleasure, and conceptually resembles the rodent intracranial self-stimulation (ICSS) paradigm (Olds and Milner, 1954). DRUG-UNRELATED STUDIES AND THEIR CONTRIBUTION TO UNDERSTANDING THE MOLECULAR BASIS OF ADDICTION: THE CASE OF LEARNING AND MEMORY Addiction is frequently referred to as pathological usurpation of learning and memory mechanisms that are normally used to predict the occurrence of natural rewards (Nestler, 2002;Hyman, 2005;Hyman et al., 2006;Kalivas and O'Brien, 2008;Duan et al., 2016;Patrono et al., 2016). This part of the review will explore the contribution of the field of learning and memory in flies to understanding drug related behaviors and possibly addiction, by covering two major directions in the field: traditional forward genetic screens, and more recent circuitry-oriented studies. Genes and Cellular Pathways That Constitute the Basic Machinery Encoding Learning and Reward Seymour Benzer and colleagues were the first to demonstrate that one can use a genetic scalpel to identify genes and pathways that are necessary for the formation of memory (Quinn et al., 1974). Learning and memory can be studied in flies using both reward or avoidance of punishment based assays, by pairing a neutral cue to the presence of sucrose (positive reinforcement) or electric shock (punishment) (Quinn et al., 1974;Tempel et al., 1983). The memory for the experience is measured by testing the avoidance or attraction of the flies to the odor that was previously paired (conditioned stimulus) with the experience (unconditioned stimulus). Many studies over the years identified mutants in different stages of the process, some of which showed virtually no learning during shock training, like the mutant turnip , dunce (Dudai et al., 1976), and rutabaga (Aceves-Piña and , while others learned normally but forgot the task of shock and sucrose training faster than wild type flies, like amnesiac Tempel et al., 1983). The elucidation of the molecular functions of the affected genes shed light on the biochemical mechanisms mediating learning and memory, and indicated a pivotal function for the cAMP pathway; rutabaga (rut) encodes for the Ca 2+ /CaM-sensitive adenylyl cyclase (Livingstone et al., 1982), and Dunce has cAMP phosphodiesterase activity (Byers et al., 1981). In addition to the cAMP pathways, other studies identified additional players that regulate memory related plasticity events, such as Ca 2+ /CaM Kinase II (Joiner and Griffith, 1997) and Orb2, a CPEB protein that functions in synaptic plasticity-required protein synthesis (Keleman et al., 2007). Studying Neuronal Circuits That Encode Associative Learning; The Mushroom Bodies as an Association Center Recent technological advances in neurogenetics led to the emergence of powerful genetic tools such as optogenetics, in vivo Ca 2+ imaging, and the ability to manipulate single neurons in behaving animals. This resulted in an explosion of studies on mechanisms that encode associative learning and the processing of natural rewards (reviewed by Owald et al., 2015). A central player in integrating the conditioned and unconditioned stimuli of a given experience into an associative memory is the Mushroom Body (MB), a brain region extensively studied with classical conditioning assays and genetic manipulations (Heisenberg et al., 1985;Connolly et al., 1996;Wolf et al., 1998;Waddell et al., 2000;McGuire et al., 2001;Liu et al., 2007;Thum et al., 2007;Aso et al., 2009) (reviewed by Kaun and Rothenfluh, 2017;Cognigni et al., 2018). Below we introduce some basic principles that govern the function of the MB, as an introduction to the neuronal machinery that processes positive reinforcement, which is required for reward learning. As such, this is not intended to be a comprehensive review of the MB [for detailed up to date reviews on the wiring and function of the MB see (Scaplen and Kaun, 2016;Felsenberg et al., 2017;Kaun and Rothenfluh, 2017;Cognigni et al., 2018)]. MBONs integrate sensory information with the valence of the experience (Hige et al., 2015), generating an association between the conditioned and unconditioned stimuli, and leading to associative memory formation. For this to happen, specific subpopulations of Dopaminergic Neurons (DAN) that innervate each compartment deliver information about the valence of the experience (unconditioned stimulus) (Thum et al., 2007;Aso et al., 2010Aso et al., , 2014aLiu et al., 2012;Caron et al., 2013;Clowney et al., 2015) (reviewed by Das et al., 2016). Activation of different populations of DANs is sufficient for aversive or appetitive memory formation when paired with a CS (reviewed by Waddell, 2013). Further functional dissections revealed that different subpopulations of DANs and MBONs encode information regarding the sweet vs. caloric value of the ingested food (Das et al., 2014), water reward (Shyu et al., 2017), aversive taste (Masek et al., 2015), electric shock memory (Unoki et al., 2005;Aso et al., 2010), and even specific short and longterm memory formation (Aso et al., 2014b). Memory formation, consolidation, retrieval, reconsolidation and/or extinction have been shown to occur via neuronal activities in specific parts of the MBONs and specific subsets of reinforcing DANs (Berry et al., 2012(Berry et al., , 2015Shuai et al., 2015;Aso and Rubin, 2016;Ichinose and Tanimoto, 2016) reviewed by Cognigni et al. (2018). Intriguingly, re-evaluation of previously learned appetitive memory was shown to be conveyed by the activity of a subset of MBONs that is anatomically connected to both aversive and appetitive DANs, and that extinction or re-consolidation of appetitive memories requires activity of both during re-evaluation (Felsenberg et al., 2017). Finally, a recent comprehensive connectome map of the entire MB alpha lobe that was generated by electron microscopy imaging, demonstrated that the interconnectivity between KCs, DANs and MBONs is even more intricate than previously thought, paving the path for further delineation of the underlying neurobiological principles of this brain region (Takemura et al., 2017). Shared Molecular Machinery of Memory, Reward, and Drug-Related Behaviors in Model Organisms Drug rewards converge on molecular and neural pathways that encode memory for natural rewards, and induce similar neuroplastic changes as natural rewards (reviewed by Hyman et al., 2006;Kauer and Malenka, 2007;Kalivas and O'Brien, 2008;Koob and Volkow, 2010). The cAMP, CREB dependent and FosB pathways play a prominent role in mediating these long-term adaptive changes in neuronal function (Nestler, 2002;Mameli and Lüscher, 2011). An example of the crosstalk between natural reward, drug reward and neuroplasticity is demonstrated in studies where periods of abstinence from sexual experience increase the sensitization of rats to amphetamine reward (Bradley and Meisel, 2001;Pitchers et al., 2010). This sex experience-induced plasticity, which in turn causes enhanced drug reward, was shown to be mediated by dopamine 1 receptor (D1R)-dependent induction of FosB in the nucleus accumbens (NAc) (Pitchers et al., 2013). A similar phenomenon was also documented in Drosophila, in which sexual deprivation increased the motivation to consume ethanol as a drug reward, by regulating the brain levels of neuropeptide F (NPF) (Shohat-Ophir et al., 2012). As stated previously, the dopaminergic system plays a central role in processing natural rewards, and represents one way by which drugs of abuse induce changes in memory-related mechanisms (Di Chiara, 1999). In mammals, dopaminergic neurons show characteristic burst-firing activity during mating and food consumption (Dackis and O'Brien, 2001). Cocaine increases dopaminergic neurotransmission by blocking dopamine transport, preventing its removal from the synaptic cleft (Dackis and O'Brien, 2001). Reducing dopamine levels in fruit flies, using a competitive agonist to tyrosine hydroxylase (which converts tyrosine to L-Dopa), diminishes their sensitivity to cocaine and nicotine (Bainton et al., 2000). Dopamine release is also required for the expression of ethanol reward in fruit flies, as temporal block of neurotransmission in dopaminergic neurons prevented conditioned preference for ethanol-associated cues (Kaun et al., 2011). In addition, artificial activation of a certain dopamine neurons such as the protocerebral anterior medial (PAM neurons) is rewarding per se, as it induces robust appetitive odor memory in the absence of natural or drug reward (Liu et al., 2012). Shared Mechanisms for Ethanol-Related Behaviors and Learning and Memory in Flies Examining the connection between neuroplasticity and drug response, several studies tested whether established learning and memory mutants also depict aberrant behavioral phenotypes in acute ethanol response. The mutant cheapdate, which is an allele of the memory mutant amnesiac, caused increased sensitivity to the sedating effects of ethanol (Moore et al., 1998;Wolf and Heberlein, 2003). Another mutant, rut, exhibited increased ethanol hyperactivity and sensitivity (Wolf et al., 2002;Heberlein et al., 2004). In addition to acute responses to ethanol, learning and memory mutants revealed altered rapid and chronic tolerance responses to ethanol (for detailed list of genes see (Berger et al., 2008). For instance, the long-term memory mutant john displayed enhanced chronic tolerance in response to prolonged exposure (20-28 h) to low concentration of ethanol vapor (Berger et al., 2008). Krasavietz (or exba), which encodes a translation initiation factor, is an example of a gene involved in learning and memory whose mutation affects both acute ethanol response and the motivation to consume ethanol. Krasavietz mutant flies exhibit decreased sensitivity to ethanol sedation (Berger et al., 2008), defects in the development of ethanol tolerance (Berger et al., 2008), and reduced voluntary consumption of ethanol (Devineni and Heberlein, 2009). Moreover, the expression of the memory gene rut in mushroom body (MB) neurons is necessary for robust ethanol consumption (Xu et al., 2012). Recent studies identified new players that connect neuroplasticity and the formation of memories to the rewarding effects of ethanol intoxication. scabrous, which encodes a fibrinogen-related peptide that regulates Notch signaling, was shown to be necessary for the rewarding effects of ethanol intoxication (Kaun et al., 2011). Another study discovered that the sirtuin gene Sir2 (Sirt1), which deacetylates histones and transcription factors, is regulated by exposure to ethanol vapor, and is required for normal ethanol sensitivity, tolerance, and for ethanol preference and reward (Engel et al., 2016). Lastly, although this review focuses on ethanol related behaviors, it is important to mention a study that tested the role of memory mutants in nicotine-induced motor sensitivity (Hou et al., 2004). Using a startle-induced climbing assay, measuring the effect of nicotine vapor on climbing ability, Hou et al. demonstrated that dunce mutant flies, which harbor higher basal levels of cAMP, exhibited increased sensitivity to the depressing effects of nicotine. In contrast, DCO H2 (Pka-C1 H2 ) and DCO B3 (Pka-C1 B3 ) mutants that are defective in PKA showed low sensitivity to nicotine (Hou et al., 2004). MOTIVATIONAL STATES AS AN ORGINIZING PRINCIPLE THAT SHAPES REWARD PROCESSING Animals continuously integrate their internal physiological state with environmental signals, and subsequently choose one action over another to increase their chances of survival and reproduction. As such, the state of the organism defines which stimuli are positively reinforced, negatively reinforced or considered negligible. A classic example of this is that fruit flies have to be hungry to express appetitive memory for sugar (Krashes and Waddell, 2008;Krashes et al., 2009;Gruber et al., 2013), highlighting the ability of internal signals such as hunger to modulate learned responses of cues associated with food. An example of the interplay between states and reward processing can also be seen in aversive conditioning in fruit flies. Pairing a neutral odor with electric shock forms an association that predicts the arrival of pain. Conversely, presenting the odor following electric shock promotes appetitive behavior, and predicts the relief of pain, implying that the end of an aversive state can also be rewarding (Tanimoto et al., 2004). This indicates that even in flies, reward is not an absolute experience, but is relative to the state in which it is perceived. Repeated stressful experiences, such as repeated exposure to heat or electric shocks, where the fly cannot evade punishment by walking away, can induce a depression-like state, leading to decline in walking activity, similar to learned helplessness paradigms in rodents (Yang et al., 2013). Uncontrollable repeated mechanical stress in flies can induce long-lasting changes in motivational states, exhibited by reduced motivation to seek rewards and reduced 5HT (serotonin) levels (Ries et al., 2017). This depression-like state can be relieved by lithium treatment or artificial activation of serotonergic neurons that project to the MB (Ries et al., 2017). Another aspect of the interplay between motivational states and reward is the concept that different motivational states are associated with particular drives (reward seeking behavior) and specific sensory sensitivity. For instance, food deprivation and satiety affect the extent of foraging behavior and food consumption, and modulate sensory perception of food related sensory stimuli (Lee and Park, 2004;Yu et al., 2004;Wu et al., 2005;Root et al., 2011;Inagaki et al., 2012Inagaki et al., , 2014bMarella et al., 2012;Beshel and Zhong, 2013;Wang et al., 2013;Ko et al., 2015;Jourjine et al., 2016). This is achieved by coordinated regulation of several different neuropeptide and hormonal systems that integrate nutrient signals and metabolic inputs into regulation of homeostatic drives and modulation of sensory systems (Lee and Park, 2004;Yu et al., 2004;Wu et al., 2005;Inagaki et al., 2012Inagaki et al., , 2014bMarella et al., 2012;Gruber et al., 2013;Wang et al., 2013;Jourjine et al., 2016) (reviewed by Landayan and Wolf, 2016). This presumably occurs via the activation of specific DAN innervating the MB. For example, it was recently shown that insulin triggers the opposing functions of two neuropeptide systems: short neuropeptide F (sNPF) and tachykinin, and this in turn regulates the sensitivity toward appetitive and aversive odors (Ko et al., 2015). Serotonergic neurons were also shown to modulate motivational states that regulate feeding behavior and sugar associated reward (Burke et al., 2012;Sitaraman et al., 2012). Recently, a set of 15 serotonergic neurons was identified, that when activated, induces a fed fly to eat as if it was food deprived, and promotes the formation of appetitive memory (Albin et al., 2015). These findings imply that specific sub-populations of neurons act to shift motivational states, and thus control the way by which sensory stimuli that is associated with the experience is processed and affects behavior. NPF System as a Molecular Signature for Reward States The NPF/NPF-receptor system is emerging as a central player in modulating and encoding motivational states associated with sugar reward, sexual, and drug reward, and the homeostatic regulation of motivational responses. The activity of NPFexpressing neurons mimics a state of food deprivation, and promotes rewarding memories in satiated flies, via a subset of downstream NPF receptor expressing dopaminergic neurons that innervate the MB (Krashes et al., 2009). Additional studies revealed NPF's role in encoding other motivational aspects of feeding, such as promoting feeding (Wu et al., 2005), encoding the valence/attractiveness of food related odors (Beshel and Zhong, 2013;Beshel et al., 2017), and enhancing sugar sensitivity in sugar-sensing sensory neurons (Inagaki et al., 2014b). In addition, NPF serves as a homeostatic integration point of two interconnected systems: sleep and feeding. NPF regulates starvation, which induces sleep suppression, suggesting that the NPF system acts to encode a hunger signal that promotes an arousal state associated with high motivation to seek food (Keene et al., 2010;He et al., 2013;Chung et al., 2017). Another example that demonstrates the interplay between motivational states and ways by which reward stimuli are perceived, is the role of NPF in integrating sexual deprivation and drug related rewards. Male flies perceive both mating interactions and ethanol intoxication as rewarding (Kaun et al., 2011;Shohat-Ophir et al., 2012). Mated male flies exhibited reduced motivation to consume ethanol containing food and have had high levels of NPF transcript, while sexually deprived male flies exhibited higher motivation to consume ethanol containing food and lower NPF transcript levels. Furthermore, activation of NPF neurons is rewarding in itself, reduces ethanol consumption, and prevents the formation of appetitive memory toward ethanol. This implies that experiences that modulate motivational states, can affect the reinforcing value of other rewarding stimuli. The causal link between environmental stimuli, NPF levels and modulation of motivational behaviors has been documented in several studies. Reduction in NPF transcription and the activity of NPF-positive neurons was observed in response to negative environmental inputs, such as the presence of parasitic wasps and sexual deprivation, while NPF induction occurred in response to mating and ethanol intoxication (Shohat-Ophir et al., 2012;Kacsoh et al., 2013;Gao et al., 2015). Altogether, this suggests that NPF neuronal systems are central to the interplay between states and reward processing. Still, further studies are required to uncover the mechanism that connect NPF neuronal activity to activity of all specific DANs that project to the MB, and the neuronal and cellular mechanisms that allow this system to represent and affect a general reward state in the brain. The different roles of NPF/R system in regulating motivational and homeostatic features of behavior are conserved between flies and mammals. A large number of studies demonstrate the central role of NPY (the mammalian homolog of NPF) in regulating feeding and the motivation to feed (Tatemoto et al., 1982;Clark et al., 1984;Flood and Morley, 1991;Kalra et al., 1997;Bannon et al., 2000;Day et al., 2005;Keen-Rhinehart and Bartness, 2007). A recent study uncovered a functional link between firing activities of NPY/AgRP neurons and energy homeostasis, wherein starvation induces an increase in NPY/AgRP firing rate, which in turn promotes re-feeding (He et al., 2016). The NPY system also functions in regulating sleep and wake homeostasis (Szentirmai and Krueger, 2006;Wiater et al., 2011;He et al., 2013). A study performed in zebrafish (Danio rerio) identified NPY signaling and NPY expressing neurons as regulators of zebrafish sleep, promoting sleep by inhibiting noradrenergic signaling, thus linking NPY signaling to an established arousal promoting system (Singh et al., 2017). In addition to its role in regulating natural physiological response, NPY has long been implicated in regulating drug addiction (for review on its role in ethanol addiction see (Thorsell and Mathé, 2017). NPY administration relieves the negative affective states of drug withdrawal and depression (Stogner and Holmes, 2000;Redrobe et al., 2002). Recently, a neuronal mechanism for the interplay between stress and reward systems on ethanol binge drinking was dissected in mice and monkeys, providing the first evidence for NPY and CRF functional interaction within neurons of the BNST (a limbic brain structure that is enriched with NPY and CRF neurons) (Pleil et al., 2015). Activation of the NPY Y1 receptor in the BNST led to enhanced inhibitory synaptic transmission in CRF neurons, which reduced binge alcohol drinking (Pleil et al., 2015). Their findings propose CRF neuronal function as a target for future therapies aimed to prevent and treat alcohol abuse. Internal state dependent modulation Sensory sensitivity (Inagaki et al., 2014b) Motivation to obtain rewards (Ries et al., 2017, Shohat-Ophir et al., 2012, Tanimoto et al., 2004 Reward value (Ries et al., 2017, Shohat-Ophir et al., 2012, Tanimoto et al., 2004 Sensitivity to sedating effects (Eddison et al., 2011) A p p e ti ti v e m e m o r ie s (K a u n e t a l. , 2 0 1 1 ) (D ev in en i et al . 20 09 ) FIGURE 2 \| Schematic model illustrating the genetic and motivational elements that influence the likelihood of progressing from initial exposure to repeated use. Red spiral depicts the multistage progression from initial drug exposure to drug dependence and addiction and the behavioral features that are shaped by molecular and neuronal mechanisms. Blue arc depicts the way by which internal state can modulate different features in reward processing via molecular and neuronal mechanisms affecting sensory sensitivity to reward related cues, the motivation to seek and obtain rewards, and the reinforcing value of the consumed reward. CONCLUDING REMARKS The risk of developing addiction is determined by molecular and neuronal mechanisms that influence the likelihood of progressing from initial drug exposure to repeated use. These mechanisms can shape the experience of initial consumption, the amount consumed, and the relative value of its reinforcing properties (Figure 2). For instance, genetic variations in bitter taste receptor and ethanol metabolism pathway influence the risk to develop addiction (Hinrichs et al., 2006;Yu and McClellan, 2016). Enhanced sensitivity to bitter taste is associated with reduced risk, and variations in ethanol metabolism lead to enhanced negative side effects and reduce the likelihood of repeated use, and therefore the risk to develop addiction (Figure 2). Studies in Drosophila demonstrated the functional link between ethanol metabolism and sensitivity to acute ethanol exposure (Ogueta et al., 2010). Other genetic components that control sensitivity to the hedonic and sedating effects of ethanol play a role in determining the extent of initial consumption and likelihood of repeated use. Upon repeated use, genetic factors that determine the extent of tolerance to ethanol-mediated responses can also shape the amount that is needed to reach the euphoric state (Figure 2). An analogy for reward states can be proposed in which high reward state is illustrated by a full "reservoir" and low state by an empty "reservoir." One can speculate that vulnerability to addiction is related to the size of "reservoir" to be filled (Bar, 2012). According to this model, bigger reservoir will require greater amounts of rewarding experiences in order to be filled. In addition, individuals can possess different sensitivity to fluctuations in the levels of reward within the reservoir, where sensitive individuals have increased motivation to fill up the reservoir with any type of reward, while others will be less affected by fluctuations, corresponding to reduced rewardseeking behavior. Lastly, prior experience/motivational states can also enter into this equation, modulating different aspects of drug response. For instance, social isolation affects sensitivity to ethanol sedation (Eddison et al., 2011), pain can modulate the perception of reward-related cues (Tanimoto et al., 2004), while sexual deprivation and stress modulate the motivation to seek and obtain rewards (Shohat-Ophir et al., 2012;Ries et al., 2017 ; Figure 2). It is postulated that these different conditions shape the repertoire and function of proteins within neurons that mediate reward processing. As a consequence, the reward baseline is shifted, which presumably modulates the motivation to obtain rewards, the value of the consumed reward, and the likelihood to continue consuming drug rewards (Figure 2). Still, the means by which different conditions and prior experiences are encoded in the reward system and lead to changes in motivational states are largely unknown. Recent advances in the ability to purify RNA from genetically tagged neuronal populations (Henry et al., 2012;Abruzzi et al., 2015), coupled with improvement in RNAseq technologies, make it now possible to bridge the gap between the specific transcriptomic repertoire and specific experiences/states. In this respect, it is now possible to profile the repertoire of coding mRNA, non-coding RNAs, and RNA modifications such as RNA editing, as well as the metabolome and proteome of specific neurons in every state. This can facilitate studies exploring the contributions of co-transcriptional mechanisms such as RNA editing, post-transcriptional mechanisms such as RNA methylation, and post-translational mechanisms in shaping the vulnerability to drug addiction. Further in-depth mechanistic studies will be required to connect specific regulation events to their functional relevance in shaping the transition from initial drug use to addiction. AUTHOR CONTRIBUTIONS All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.	2018-04-24T13:08:11.039Z	2018-04-24T00:00:00.000	{ "year": 2018, "sha1": "00045db196f63997daf119e863cd12f40adf3c59", "oa_license": "CCBY", "oa_url": "https://www.frontiersin.org/articles/10.3389/fphys.2018.00327/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "00045db196f63997daf119e863cd12f40adf3c59", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Psychology", "Medicine" ] }
51789491	pes2o/s2orc	v3-fos-license	A Novel Therapy for Melanoma and Prostate Cancer Using a Non-Replicating Sendai Virus Particle (HVJ-E) The earliest virotherapies involved injection of wild-type viruses and evaluation of their efficacies [1-3]. Ex vivo therapies using autologous irradiated tumors infected with oncolytic viruses were also investigated [12-18]. Deletion mutants of oncolytic viruses [19-24], and recombinant viruses carrying a therapeutic gene [25-32] that induce cancer apoptosis or cancer immunity have been developed and evaluated in the clinical setting. Oncolytic viruses derived from adenovirus, poxvirus, reovirus, picornavirus, paramyxovirus, and herpes simplex virus are currently available and have been clinically evaluated [33][34][35].In China, two adenovirus-based products (Gendicine and Ocorine) have been commercialized [36], while randomized phase III studies of two oncolytic viruses (reovirus and poxvirus) are ongoing in advanced countries [33][34][35].Thus, virotherapy is expected to become available as a new approach for cancer treatment, and specific product approval is anticipated in the US, EU, and Japan [37]. The major drawback associated with virotherapy is safety since replicating viruses are used in this therapy.In order to reduce the toxicity to normal cells, oncolytic viruses with strict specificity for cancer cells were constructed [29,[38][39][40][41][42].However, the use of these viruses is still considered to be high risk because it is theoretically possible that a virulent infection may occur after recombination with wild-type viruses [43]. An alternative option to avoid such a risk is to use a non-replicating oncolytic virus [44].We found that a non-replicating oncolytic virus (HVJ-E: hemagglutinating virus of Japanenvelope) is able to induce cancer cell-specific apoptosis and immunity [45].The induction of apoptosis and activation of dendritic cells in vitro, and anti-tumor activity in vivo are similar to the wild-type hemagglutinating virus of Japan (also known as Sendai virus, HVJ) [45]. The hemagglutinating virus of Japan was discovered in Sendai, Japan, in the 1950s [46].It is a paramyxovirus with a minus-strand RNA genome.The virus has fusogenic activity [47,48], and is used to prepare hybridoma cells for the production of monoclonal antibodies, and heterokaryons for chromosome analysis [49][50][51]. The hemagglutinating virus of Japan-envelope is an inactivated HVJ particle [52].It is manufactured by a process similar to that used for whole virus particle vaccines.Good manufacturing practice (GMP)-regulated processes have been established in their production for use in preclinical and clinical studies [53]. We conducted dose-setting efficacy studies for HVJ-E in a murine cancer model in which dosedependent anti-cancer activity was observed.We also conducted safety studies following good laboratory practices (GLP), including pharmacological safety studies and toxicokinetic (TK) studies in rats and monkeys, as part of an investigational new drug (IND) application. Osaka University Hospital is currently conducting two investigational clinical studies with HVJ-E for the treatment of advanced melanoma and castration-resistant prostate cancer (CRPC) [54][55][56].These clinical trials are the first human studies for HVJ-E and will reveal the safety and efficacy of the non-replicating virus (HVJ-E).Virotherapy with a non-replicating oncolytic virus is a new approach that is anticipated to provide a new strategy for cancer therapy. A new strategy for cancer therapy Most cancers are still incurable and new approaches are required to improve the efficacy of cancer treatments.However, conventional cancer therapies are problematic. Chemotherapy with anti-cancer agents is useful in achieving tumor regression.However, the immune system, which is important in the removal of residual cancer cells, is also suppressed by these agents (Figure 1).Therefore, surviving cancer cells and cancer stem cells (CSC) eventually acquire drug resistance, resulting in tumor relapse (Figure 1) [57].Thus, chemotherapy with cytotoxic drugs does not generally result in the necessary eradication of cancer cells required for long-term survival. Immune therapies for cancer offer a new approach to cancer treatment, and several products, including sipuleucel-T, are currently approved in advanced countries [58,59].The aim of these therapies is the removal of cancers by the immune system.Numerous cancer immune therapies are currently under evaluation in clinical studies.However, these agents are not potent because of lack of cytotoxic effect on cancer cells (Figure 1).These observations suggest that a two-step strategy is necessary for the eradication of cancer cells (Figure 1). During the first step, the direct killing of cancer cells is necessary for reduction of tumor volume.CSCs are usually resistant to conventional anti-cancer drugs and continue to proliferate during chemotherapy.Therefore, an agent that targets and kills CSCs is required for effective cancer treatment. During the second step, the removal of residual cancer cells (and CSCs) from the body by a cancer-specific immune response is necessary to avoid relapse of the condition [57].However, it is difficult for immune cells to recognize and remove CSCs because they exist as a minority population within the tumor, and possess a lower antigenicity than differentiated cancer cells [57].Oncolytic viruses have the capability of both directly killing cancer cells and inducing cancer immunity. Non-replicating virus particles as anti-cancer agents The use of non-replicating virus particles is a new approach in virotherapy. Kurooka and Kaneda discovered that the HVJ-E particle itself displayed anti-cancer effects in a murine model of colon cancer [45].Similar to the live (replicating) virus, HVJ-E induced maturation and differentiation of human and murine dendritic cells (DCs).It also induced infiltration of immune cells into the tumor tissue followed by activation of cancer cell-specific cytotoxic T cells.Furthermore, HVJ-E suppressed the function of regulatory T cells (Treg), which have been reported to be negative regulators of cancer immunity.Thus, HVJ-E activates cancer immunity, and simultaneously suppresses Treg [45].Fujiwara and Kaneda et al. reported that HVJ-E induced innate immunity [76].Intratumoral injection of HVJ-E promoted infiltration and activation of natural killer (NK) cells by the induction of C-X-C motif chemokine 10 (CXCL10) and type I interferons.When HVJ-E was injected into the tumor of a murine model of renal cell carcinoma (RCC), NK cells exhibited cytotoxic activity against the RCC cell line in vivo [76].The involvement of NK cells in the antitumor effect was also confirmed by showing the depletion of NK cells using an asialo-GM1 antibody [76].Activated NK cells produced interferon-γ, which induces cancer-specific cytotoxic T cells [76].These results indicated that HVJ-E is able to induce both innate and adaptive immunities. In addition to the induction of cancer immunities, HVJ-E has the capability to induce cancer cell-specific apoptosis.Kawaguchi and Kaneda et al. reported that HVJ-E showed a dosedependent, direct killing effect on human prostate cancer cell lines.In contrast, it showed no suppression of normal human prostate epithelium proliferation [77]. HVJ-E also induced apoptosis of prostate cancer cells in vivo because it showed an anti-tumor effect in severe combined immunodeficiency (SCID) mice that lack B lymphocyte-and T lymphocyte-mediated immunities [76,77].When NK cells were depleted from SCID mice by injection of the asialo-GM1 antibody, intratumoral injection of HVJ-E still showed an antitumor effect in a murine model of CRPC [77].This result suggested that HVJ-E showed a direct killing activity in vivo.HVJ-E-mediated apoptosis of cancer cells was further confirmed in a murine model of prostate cancer using a NOD/SCID mouse, which lacks both innate (NK cellmediated) and adaptive (antibody and CTL-mediated) immunities [54]. In contrast, previous studies have reported that the non-replicating oncolytic Newcastle disease virus (NDV) failed to show an anti-tumor effect in vivo [80,81].These results are inconsistent with results obtained with HVJ-E [45,76,77,79], and the putative reason is the difference between the number of particles used in the studies.In studies with NDV, 5 × 10 9 PFU of oncolytic virus were systemically or locally administrated [81], whereas the number of HVJ-E particles administered was estimated to be higher. Another possibility is the difference in the capability of the virus to deliver its RNA fragments to target cancer cells.The Z strain-derived HVJ-E used in our studies has the highest level of membrane fusion activity [52].Therefore, it is possible that HVJ-E has the ability to deliver more RNA component to the target cancer cells than the UV-inactivated NDV particle.The difference in process of inactivation may be responsible for the activity of non-replicating oncolytic viruses.The Newcastle disease virus was inactivated by UV irradiation, whereas HVJ-E was inactivated by a combination of treatment with an alkylating agent and UV irradiation [53].Inactivation conditions affect the efficiency of delivery, and strictly regulated processes are necessary to obtain suitable performance [53]. Treatment of cancer cells with HVJ-E enhanced the expression and activation (cleavage) of caspases 3, 8, and 9 (Figure 4A) [77], and induced dose-dependent apoptosis of melanoma, prostate, and other cancer cell lines in vitro (Figure 4B) [77].Interestingly, no apoptotic effects were observed on normal epithelial cells derived from murine prostate [54,77].Thus, the apoptotic activity of HVJ-E is considered to be specific to cancer cells [54,77].Matsushima and Kaneda et al. conducted an investigation to determine the active component for anti-cancer effects, and identified RNA fragments within the particle [54].Moreover, Kaneda et al. analyzed the signaling pathway involved and revealed that retinoic acid inducible gene-I (RIG-I) is a key factor for signal transduction [44,54,77].Retinoic acid inducible gene-I is a cytosolic nucleic acid receptor and was originally identified as a sensor that recognizes infection by single strand RNA viruses [82,83].Thus, RIG-I has been recognized as an inducer of immune response against infected viruses [82][83][84]. The RNA fragments delivered by HVJ-E bind the helicase domain of RIG-I in the cytoplasm and change the conformation to unmask the caspase activation and recruitment domain (CARD) (Figure 3B).After binding with the RNA fragments, RIG-I interacts with the mitochondrial antiviral signaling (MAVS) protein on the mitochondrial membrane (Figure 3B) [84]. It has been reported that the RIG-I/MAVS pathway is activated after cancer cells are treated with HVJ-E [54].Transfection of isolated RNA fragments from HVJ-E particle also induced apoptosis of cancer cells in vitro [54].Thus, HVJ-E uses a natural oligonucleotide (RNA fragment from the inactivated virus genome) as an active ingredient, and a natural virus particle (envelope of HVJ) as a delivery system for a nucleic acid medicine (RNA fragments). The apoptosis induced by HVJ-E was cancer cell-specific because normal endothelial cells showed no apoptosis after the treatment with HVJ-E [54,77]. Involvement of RIG-I in cancer cell-specific apoptosis has also been indicated in ovarian cancers and melanoma.The authors used pppRNA and poly(I:C) as ligands for RIG-I and MDA-5, and showed the involvement of caspase-9 and Apaf-1 during apoptosis.They also reported the reduction of lung metastasis by treatment with ligands for RIG-I and MDA-5 in the NOD/SCID mouse [90].Details of the underlying pathways are currently being analyzed using siRNAs of apoptosis-related factors [54].It has been suggested that differences in the expression of apoptotic genes such as Noxa, TRAIL, and TRAIL receptors in cancer cells and normal cells determine the specificity of apoptosis induced by HVJ-E (Figure 3B) [54]. Efficacy in preclinical studies The efficacy of the non-replicating oncolytic virus (HVJ-E/GEN0101) was examined in murine models of melanoma and prostate cancer (Figure 5A).GEN0101 is the identification code for the agent.The data have indicated that the anti-tumor effect is dose-depend-ent similar to other non-viral anti-cancer agents (Figure 5B).This feature is important for the development of non-replicating oncolytic virus as a novel therapeutic agent; identification of the optimal dose for non-replicating oncolytic viruses is easier than determining the optimal dose for replicating oncolytic viruses as the latter are subject to change (increase) resulting from replication in target cancer cells.The effects of administration in a xenograft model of human CRPC were also examined (Figure 5A).Efficacy after subcutaneous administration was revealed (Figure 5C).It is known that the SCID mice lacks B lymphocyte-and T lymphocyte-mediated immunities such as antibody production, and induction of cytotoxic T cells but retains monocytes and NK cells important for innate immunity.Thus, the non-replicating virus (HVJ-E/GEN0101) is able to induce innate immunity, and show anti-cancer activity in vivo even in the absence of direct killing of cancer cells.Efficacy by subcutaneous administration is important for the development of a non-replicating virus because subcutaneous administration is more common than intratumoral administration.Intratumoral administration kills cancer cells directly and promotes the release of tumor antigens, which are recognized by immune cells.Subcutaneous administration is expected to enhance and sustain the immune response.Therefore, the combination of intratumoral and subcutaneous administration is suggested as a suitable regimen in the clinical setting. In summary, the results of efficacy studies have indicated that the anti-cancer effect of the non-replicating oncolytic virus HVJ-E/GEN0101 is dose-dependent similar to conventional anti-cancer drugs, and subcutaneous administration may be the preferred administration route for the viral and non-replicating viral agents. Safety studies Safety of the non-replicating oncolytic virus (HVJ-E/GEN0101) has been confirmed in nonprimate (rat) and primate (Cynomolgus monkey) animals.Lists of the studies that have been conducted are shown in Table 1A and B. Results from single dose general toxicity studies revealed that no death or severe finding was observed even in the maximum dosage groups.Similar to the single dose studies, no severe finding was observed in repeated dose, general toxicity studies (Table 1A). Results from immunological and genetic toxicity studies in rat and monkeys revealed that no abnormal symptoms related to the test agent were observed.The levels of IL-6 and IFN-γ in monkey serum were analyzed after subcutaneous injection of HVJ-E/GEN0101, and the levels of both cytokines were determined to be within the normal range (data not shown).A core battery of safety pharmaceutical studies was performed to determine the effects on major organs (central nervous, respiratory, and cardiovascular systems); no abnormal effect was observed with the exception of transient and non-severe pyrexia (Table 1B).for the efficacy study using murine models of melanoma and prostate cancer.Left: C57/BL6 mice were transplanted with B16/BL6 cells and 3 intratumoral injections of HVJ-E/GEN0101 were administered after tumor formation (4 days after transplant).A time course study of change in tumor volume was performed, and the difference in tumor volume among the three groups was statistically analyzed at the end of the study.Significant differences were determined by Dunnett's multigroup test, and significant differences to the medium (control) group were observed (p < 0.01).The survival rate after the injection of HVJ-E/GEN0101 was also monitored in the melanoma model.A survival study was performed and the difference among the three groups was statistically analyzed at the end of the study.Significant differences were determined by log-rank analysis.Significant differences to the medium group were observed (p < 0.001).Right: Protocol for the efficacy study in a murine model of prostate cancer.Human CRPC (PC-3)-bearing mice were used for the study.A summary of the study protocol is shown.The anti-tumor effect of HVJ-E/GEN0101 for each administration route was examined.Severe combined immunodeficiency mice were transplanted with PC-3 cells.Two intratumoral injections or 6 subcutaneous injections of HVJ-E/GEN0101 were performed after tumor formation (4 days after transplant).A time course study of change in tumor volume was performed and differences in tumor volume between the two groups were statistically analyzed at the end of study.Significant differences were determined by t-test and significant differences to the medium were observed in both routes (p < 0.01 and p < 0.05).(B) Efficacy study in a murine model of melanoma.Dose-dependency of the anti-tumor effect of HVJ-E/GEN0101 was revealed. (C) Efficacy study in a murine model of prostate cancer.A time course study of change in tumor volume was performed and differences in tumor volume between the two groups were statistically analyzed at the end of the study.Significant differences were determined by t-test, and significant differences to the medium were observed for both administration routes (p < 0.01 and p < 0.05). Clinical studies Two clinical studies using the non-replicating oncolytic virus (HVJ-E/GEN0101) are currently being conducted in Osaka University Hospital.The target diseases are advanced melanoma (stage IIIC and stage IV) and CRPC [55,56].These proof-of-concept studies in melanoma and CRPC using the non-replicating oncolytic virus were initiated in July 2009 and July 2011, respectively [55,56].The respective summaries of both studies are shown in Table 2A and B. The primary endpoints of these studies were safety and tolerability based on the Common Terminology Criteria for Adverse Events (CTCAE) version 4.0, whereas the secondary endpoints were efficacy and confirmation of the mode of action.The major difference between the regimens for the melanoma and CRPC studies was the route of administration and number of administrations.A combination of intratumoral and subcutaneous routes of administration (one intratumoral and three subcutaneous injections) was adopted in the CRPC study.In addition, a new injection system developed by Okayama University was used in the CRPC study.This system permits stable refrigerated storage of the test article, and accurate injection into the prostate [91]. Discussion and conclusion The major disadvantages associated with oncolytic viruses are safety concerns because viral replication could theoretically cause the emergence of new pathogenic viruses [43].The use of non-replicating oncolytic viruses is expected to resolve the safety issues associated with conventional oncolytic viruses because they are unable to replicate in target cells. Novel Gene Therapy Approaches The modes of action underlying the anti-cancer effects of non-replicating virus on cancer cells and immune cells have been analyzed.One major signaling pathway is the RIG-I/ MAVS pathway [54].RIG-I is a cytosolic nucleic acid receptor that acts as a sensor that detects virus infection [82][83][84].Similar to wild-type RNA viruses, the non-replicating virus (HVJ-E) is able to activate the RIG-I/MAVS pathway in DCs and induce both innate and adapted immunities.The non-replicating virus also activates the RIG-I/MAVS pathway in cancer cells and induces cancer cell-specific apoptosis.Genetic analyses suggest that differences in the expression of apoptosis-related genes define the sensitivity to the treatment with a non-replicating virus (HVJ-E) [54].Furthermore, it is suggested that methylation of the respective enhancer/promoter regions underlies differences in transcription of apoptosis-related genes [54]. RIG-I, TLR3, and MDA-5 signaling pathways are involved in the apoptosis of ovarian cancers [87][88][89] and melanoma [90].Thus, the RIG-I/MAVS pathway is likely to emerge as a new target for the development of drugs that induce cancer cell-specific apoptosis. The non-replicating virus (HVJ-E) activates DCs to produce IL-6, which suppress the function of Treg [45].This effect is expected to maintain the induced cancer immunity because Treg is known to be a negative regulator of immune responses [92][93][94].It has been reported that cancer cells escape cancer immunity by the recruitment and activation of Treg.Therefore, it will be important to control the function of Treg for long-term effective induction and maintenance of cancer immunities.Suzuki and Kaneda et al. reported that the RIG-I/MAVS pathway was not required to induce the expression of IL-6 [95].The attachment of the HVJ-E particle to the surface of DCs was sufficient for the production of IL-6, suggesting that the RNA fragments are unnecessary for the induction of this cytokine [95].Detailed analyses identified that the F protein on the surface of HVJ-E is involved in the production of IL-6 [95].Binding of the F protein to target cells requires expression of the HN protein [96].Several gangliosides, such as GD1a and sialyl paragloboside, are implicated in the association of the HVJ-E particle and cancer cells because the HN protein binds the sialic acids of gangliosides.The receptor for the F protein remains unidentified to date.Taken together, the RIG-I/MAVS signal pathway, and a second pathway that induces the production of IL-6 may cooperate in the activation and sustainment of cancer immunity induced by the non-replicating virus (HVJ-E). The development of a non-replicating oncolytic virus other than HVJ-E is possible because the manufacturing process for such a particle is similar to that of whole particle viral vaccines.In case of HVJ-E, the virus is inactivated by treatment with an alkylating agent and UV irradiation, a process used for the production of vaccines against viral diseases.Thus, the development of oncolytic viruses could be converted to the development of non-replicating oncolytic particles by similar manufacturing processes. A disadvantage associated with non-replicating oncolytic viruses may be the defect in transmission ability.Therefore, it is possible that a greater amount of non-replicating oncolytic virus may be required for effective treatment compared with a live oncolytic virus.Alternatively, more frequent injections may be necessary for complete tumor eradication compared with the use of live oncolytic viruses.However, it is important to achieve a balance between the risks and benefits associated with the therapy.In our opinion, repeated administration of the non-replicating virus should be tolerable because no severe finding was observed during our safety studies. In conclusion, non-replicating virus particles such as HVJ-E may resolve the safety issue of conventional virotherapy and provide a new strategy in cancer treatment. Future perspectives The first non-replicating oncolytic virus (HVJ-E) is currently under evaluation in clinical studies.Proof-of-concept data for non-replicating viruses in both clinical and non-clinical studies are necessary for further development of this approach.Osaka University Hospital is currently conducting two phase I/IIa studies: one for advanced melanoma and another for CRPC [55,56].The results of these studies will reveal the safety, efficacy, and optimal dosage regimen necessary for phase II study or randomized, double blind phase III study. Combination treatment may be an effective approach to increase efficacy [97].Indeed, an increase in therapeutic efficacy has been reported for virotherapies combined with photodynamic therapy [98,99], radiotherapy [100], chemotherapy [101,102], or gene therapy [103].Kiyohara and Kaneda reported that combination of the non-replicating virus (HVJ-E) and gene therapy (IL-12) increased efficacy in a murine model of melanoma [104].Furthermore, it was reported that a combination of non-replicating virus (HVJ-E) and chemotherapy [bleomycin or cis-diamminedichloroplatinum (CDDP)] increased efficacy in murine models of colon and bladder cancers [78,105]. Technologies for systemic administration and targeting for HVJ-E are under development.The HN protein of HVJ-E has hemagglutinating activity and causes agglutination and lysis of erythrocytes in vitro.Currently, inactivation of the HN protein, decreased expression of the HN protein, and "masking" with platelets is being developed for intravenous injection of HVJ-E.Targeting after the intravenous injection is also important for systemic delivery.The addition of transferrin, a single chain antibody, or platelets have been suggested as suitable modifiers for HVJ-E. The selection of viruses, or viral strains for the preparation of non-replicating oncolytic viruses is also important for obtaining higher efficacy because the level of immune response is dependent on the selection of virus strains [106].A number of replicating oncolytic viruses are currently under clinical development [35,97].Therefore, it may be feasible to select a suitable virus for therapeutic application from the oncolytic viruses that have been developed [106,107].The tropism and complement resistance features of each virus should be considered for targeting and stabilization in serum. HVJ-E is the only non-replicating oncolytic virus currently undergoing clinical investigation.These studies establish a new strategy for the virotherapy and gene therapy fields.The primary goal is to provide a novel approach for improving cancer therapy. Figure 1 . Figure 1.Problems with conventional therapies and a 2-step strategy for cancer treatment.Conventional anticancer therapies are problematic and a 2-step therapeutic strategy is proposed for effective cancer treatment.Virotherapy possesses ideal characteristics for such a 2-step therapy.Problems associated with conventional anti-cancer drugs and immune therapies, including cancer vaccines, are shown. adapted to GMP guidelines, and pilot plant for clinical trial is available b. Freeze-dried formulation is stable for over 21 months in refrigerator. Figure 2 . Figure 2. Characteristics and structure of HVJ-E/GEN0101.The characteristics of HVJ-E/GEN0101 are shown on the left.A schematic structure of the particle is shown on the right. Figure 3 . Figure 3.The mechanism of anti-cancer effects of HVJ-E/GEN0101.A novel type of virotherapy agent (HVJ-E/ GEN0101) has a multi-mode of action that is ideal for 2-step cancer treatment.(A) Target cells and sequential anticancer effects of HVJ-E/GEN0101.(B) Signaling pathway induced by stimulation with HVJ-E/GEN0101.The RIG-I/ MAVS pathway is the major pathway involved.RIG-I is a cytosolic receptor for nucleic acids: it usually functions as a sensor to recognize viral infection.The nucleic acids in the HVJ-E/GEN0101 particle act as an agonist for RIG-I and induce cancer cell-specific gene expression followed by the induction of apoptosis. Figure 4 . Figure 4. HVJ-E/GEN0101 induces apoptosis of human and murine melanoma cells.(A)Induction of caspase activity after treatment of melanoma cells with HVJ-E/GEN0101.Human (A375) and murine (B16/BL6) melanoma cells were treated with various amounts of HVJ-E/GEN0101, and caspase activities were measured 24 hours later.A dosedependent activation was observed.(B) Survival of melanoma cells after treatment with various amounts of HVJ-E/ GEN0101.Human (A375) and murine (B16/BL6) melanoma cells were treated with various amounts of HVJ-E/ GEN0101 and cell survival was measured by WST-8 assay 72 hours later. * 5 DayFigure 5 . Figure 5. Efficacy studies of HVJ-E/GEN0101 in a murine model of melanoma and prostate cancer.(A)Protocolsfor the efficacy study using murine models of melanoma and prostate cancer.Left: C57/BL6 mice were transplanted with B16/BL6 cells and 3 intratumoral injections of HVJ-E/GEN0101 were administered after tumor formation (4 days after transplant).A time course study of change in tumor volume was performed, and the difference in tumor volume among the three groups was statistically analyzed at the end of the study.Significant differences were determined by Dunnett's multigroup test, and significant differences to the medium (control) group were observed (p < 0.01).The survival rate after the injection of HVJ-E/GEN0101 was also monitored in the melanoma model.A survival study was performed and the difference among the three groups was statistically analyzed at the end of the study.Significant differences were determined by log-rank analysis.Significant differences to the medium group were observed (p < 0.001).Right: Protocol for the efficacy study in a murine model of prostate cancer.Human CRPC (PC-3)-bearing mice were used for the study.A summary of the study protocol is shown.The anti-tumor effect of HVJ-E/GEN0101 for each administration route was examined.Severe combined immunodeficiency mice were transplanted with PC-3 cells.Two intratumoral injections or 6 subcutaneous injections of HVJ-E/GEN0101 were performed after tumor formation(4 [89]88]and Barchet et al.reported that the RIG-I agonist (Poly(dAdT)) induced apoptosis and expression of MHC class I molecules in ovarian cancer cells[87,88].Van and Bell et al.also reported apoptosis of ovarian cancer cells after treatment with dsRNA[89].Analysis by shRNA-mediated knockdown revealed that RIG-I and other receptors for dsRNA (TLR3 and MDA-5) were involved in the caspase 8/9- [90]ated apoptosis of cancer cells.Similar to sensitivity of HVJ-E, epithelial cells derived from ovarian surface was resistant to apoptosis mediated by RIG-I signal pathway.When combined with conventional chemotherapy (carboplatin/paclitaxel), treatment with dsRNA showed a synergistic suppression of ovarian cancer cell viability[89].Besch et al. reported that stimulation of RIG-I and MDA-5 induced apoptosis of human melanoma cells[90]. Table 1 . Summary of toxicological studies Table 2 . Design of investigational clinical studies	2017-09-17T22:32:48.441Z	2013-02-13T00:00:00.000	{ "year": 2013, "sha1": "c7f698472b5cb742e6155fec56409047fc1c62a8", "oa_license": "CCBY", "oa_url": "https://www.intechopen.com/citation-pdf-url/42757", "oa_status": "HYBRID", "pdf_src": "MergedPDFExtraction", "pdf_hash": "1824d86090f0df152a3ee7fc03ed031bdf318670", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Biology" ] }
227249452	pes2o/s2orc	v3-fos-license	Body Localization of ACE-2: On the Trail of the Keyhole of SARS-CoV-2 The explosion of the new coronavirus (SARS-CoV-2) pandemic has brought the role of the angiotensin converting enzyme 2 (ACE2) back into the scientific limelight. Since SARS-CoV-2 must bind the ACE2 for entering the host cells in humans, its expression and body localization are critical to track the potential target organ of this infection and to outline disease progression and clinical outcomes. Here, we mapped the physiological body distribution, expression, and activities of ACE2 and discussed its potential correlations and mutal interactions with the disparate symptoms present in SARS-CoV-2 patients at the level of different organs. We highlighted that despite during SARS-CoV-2 infection ACE2-expressing organs may become direct targets, leading to severe pathological manifestations, and subsequent multiple organ failures, the exact mechanism and the potential interactions through which ACE2 acts in these organs is still heavily debated. Further scientific efforts, also considering a personalized approach aimed to consider specific patient differences in the mutual interactions ACE2-SARS-CoV-2 and the long-term health effects associated with COVID-19 are currently mandatory. SARS-CoV-2 Clinical Characteristics Since its discovery in December 2019 the coronavirus disease , caused by the transmission of a novel coronavirus known as SARS-CoV-2 induced pneumonia, infected more than 37,800,000 people worldwide and caused more than 1,080,000 deaths until October, 2020. COVID-19 patients mainly displayed pneumonia-associated symptoms, such as fever, shortness of breath, cough, sputum production, and myalgia or fatigue (1,2). However, despite SARS-CoV-2 infection is manifested as a respiratory tract infection, it may causes symptoms associated to multiple organs, including intestine and stomach (diarrhea, anorexia, nausea, vomiting, and abdominal pain), liver (abnormal enzymes levels), pancreas (pancreatitis), kidney (protein and blood in their urine, abnormal creatinine level), brain (strokes, seizures, confusion, and brain inflammation), heart and blood vessels (elevations of cardiac injury biomarkers, palmus, chest distress, cardiac inflammation and injury, arrhythmias, and blood clots), eyes (conjunctivitis, inflammation of the membrane that lines the front of the eye, and inner eyelid), nose (anosmia), ect (3)(4)(5)(6)(7)(8)(9)(10)(11). This multiple organ involvement can lead to a poorer outcome to the viral infection and often result in hospitalization and intensive care unit (ICU) admittance (12)(13)(14). Despite the mechanisms for high morbidity and mortality induced by SARS-CoV-2 are currently unknown, based on available literature data in public databases, it is known that the risk of infection and mortality increases with advancing age and also seems to show a sexual dimorphism, male elderly subjects are at higher risk of infection, as well as death (1,2). In addition, despite COVID-19 is a non-discriminatory disease, involving both healthy individuals and those with comorbidity conditions, it is well-documented that mortality further increases in presence of pre-existent pathologies, such as cardiovascular disease, hypertension, diabetes, obesity, chronic pulmonary disease, and cancer (12)(13)(14). Despite, the biological mechanisms behind these observations are still unclear, virus/host cell interaction, immunological differences, and sex-based hormonal differences are likely to be involved. Interaction Between SARS-CoV-2 and ACE2 Mechanisms implicated in SARS-CoV-2/host cell interaction are of key importance for cell infection and replication that in turn lead to disease and related damage. In this context, the angiotensin converting enzyme 2 (ACE2), an enzyme important in the renin-angiotensin-aldosterone system (RAAS), scarcely present in the circulation, but widely expressed in organs and able to regulate blood pressure and fluid balance, has been seen to play a key role (15,16). ACE2 operates as an ACE counterpart: it acts as a carboxypeptidase, removing single amino acids, converting Ang II to its metabolite angiotensin-(1-7) (Ang1-7), balancing the effects of Ang II. ACE2 is found to the apical surface of epithelial cells, differently from ACE, which is located between the apical and basolateral membranes in polarized cells. ACE2 plays its pivotal role in regulating blood pressure and consequently hypertension. This activity is mediated by the ACE2/Ang-(1-7)/Mas receptor axis, through which the regulation of angiotensin and Ang-(1-7) and nitric oxide (NO) availability control blood pressure alterations, which cause damages to vascular tissue as atherosclerosis, hypertrophy and more in general, endothelial alterations (17). Operatively, there are two forms of ACE2: (1) the full-length ACE2, that presents a structural transmembrane domain able to anchor its extracellular domain to the plasma membrane, and (2) the soluble form of ACE2, that lacks the membrane anchor and circulates in small amounts in the blood (15,16). SARS-CoV-2 enters cell by the binding of spike (S) viral protein, an amino acid long protein that belongs to the viral envelope and leans outwards with a "corona" like form, to the ACE2 receptor (16,18). The initial step of viral entry is represented by the binding of the N-terminal domain of the viral protein unit S1 to a pocket of the ACE2 receptor. After this, the receptor transmembrane protease serine 2 (TMPRSS2), a member of the Hepsin/TMPRSS subfamily that is stechiometrically contiguous to ACE2 receptor, induces the cleavage of the protein between the S1 and S2 units, with the help of Furin which facilitates the entry of the virus into the cell after binding (19,20). Furin [also termed paired basic amino acid cleaving enzyme (PACE)], a member of the subtilisinlike proprotein convertase family that processes protein of the secretory pathway, is expressed in multiple organs, such as in lungs, liver, and small intestines. Following the binding of the S glycoprotein to ACE2, furin-mediated proteolytic cut of the S protein is necessary for viral entry into the cell (18,21). Thus, both TMPRSS2 and Furin are crucial for S activation. The key role of these two proteases was also demonstrated by a recent study that showed that multicycle replication of SARS-CoV-2 in Calu-3 human airway cells was strongly suppressed by inhibiting TMPRSS2 and Furin activity (22). However, virtually, other human proteases, e.g., cathepsin L and B, elastase, trypsin and factor X, may be involved in the entry of SARS-CoV-2 into the human cell and in the shedding of ACE2. A critical cell membrane protease involved in the endogenous shedding of ACE2 from membranes is the disintegrin metalloproteinase 17 (ADAM17), also known as tumor necrosis factor-α converting enzyme (TACE) (23). While TMPRSS2 cleaves both ACE2 and the S protein of SARS-CoV-2, leading to membrane fusion and cellular uptake of the virus, ADAM17 acts directly and solely on ACE2 and leads to ACE2 shedding into the extracellular cellular space. Thus, ADAM17 and TMPRSS2 may have opposite effects on ACE2 shedding. Evidences have shown that the expression of TMPRSS2 inhibits ADAM17-shedding of ACE2 (24). However, it is unclear how TMPRSS2 transcends ADAM17 to cleave ACE2 during SARS-CoV-2 infection. Despite numerous information has been obtained up to now, the exact role of ACE2 in SARS-CoV-2 cellular infection and of proteases that process SARS-CoV-2 S protein is not yet defined. Certainly, genetics and demographic characteristics, lifestyle, comorbidities, and medication usage may have an impact on ACE2 expression and activity in SARS-CoV-2 cellular infection. Risk Factors for COVID-19 Severity and ACE2 Expression ACE2 is regulated by a gene which maps on the X chromosome (Xp22.2), thus suggesting that some differences may exist in the expression of ACE2 between men and women (25). In women to prevent the redundant expression of the products of the genes present in double copy on the X chromosomes, a physiological random inactivation occurs in one of the two chromosomes (25). The remained chromosomal portions that escape to the inactivation and the genes present in these areas (∼15%) can be over-expressed in women (25). ACE2 is encoded precisely in these regions of the X chromosome which escape the inactivation of one of the two X chromosomes, supporting the hypothesis of a greater ACE2 expression in women (25). There is evidence that ACE2 tissue levels are also regulated by estrogens that can increase the presence of ACE2 receptor (26). Thus, if, as reported by several commentary in literature, the presence of ACE2 throughout the body could make tissues more vulnerable to SARS-Cov-2 infection women should be more predisposed to the virus than men (26). On the contrary, epidemiological data of the World Health Organization (WHO) highlighted gender-based clinical differences in SARS-CoV-2, with a higher mortality rates in male patients, in particular elderly patients (27). Even this latest information appears to be in contrast with the hypothesis that ACE2 throughout the body could make tissues more vulnerable to SARS-Cov-2 infection. In fact, it was demonstrated that ACE2 level decrease with age and seem to be higher in young people that commonly develop a less severe COVID-19 form (26). It is important to underline that also the opposed hypothesis, that a mild/moderate ACE2 deficiency may protect from SARS-CoV-2 invasion, seems improbable considering the high affinity of the virus for ACE2 receptor. In addition, this latter hypothesis is also unlikely because different degree of ACE2 deficiency are related with specific diseases, i.e., diabetes, obesity and cardiovascular disease, that characterize individuals more prone to be infected and to have severe complications related to SARS-CoV-2. These inconsistencies highlight that other factors, such as for example organ-specific ACE2 distribution and expression levels and potential co-expression and interaction with specific proteases, may contribute to the severity of SARS-CoV-2. Although it is demonstrated that lungs inflammation is one of the main symptom during SARS-CoV-2 infection, the lungs, among all organs, present a moderate expression of ACE2 and, as reported above, SARS-CoV-2 may affect other organs, organs that have a high to moderate expression of ACE2. In this context a detailed map of the physiological organ-specific distribution, expression, and activities of ACE2, also considering organ-specific gender biases and organs often poorly considered (specific brain regions, oral cavity, thyroid, pancreas, duodenum, colon, rectum, gallbladder, male -testis and seminal vesicle-and female tissues -ovary, oocyte, uterus, vagina-, skin, and others), and a complete overview on the potential link between these organs and SARS-CoV-2 may contribute to understand the potential infection routes as well as the clinical symptoms and mechanisms of the virus susceptibility. ACE2 IN HUMAN PHYSIOLOGY: BODY LOCALIZATION, EXPRESSION, FUNCTION AND ACTIVITIES About 20 years ago, the first paper reported the mapping of ACE2 in 72 tissues (28). Over the years, it has become more and more clear that ACE2 localization can be quite tricky (28). Starting from the localization in the renal and cardiovascular tissues, over time it has become evident that ACE2 is also present in tissues and organs where initially no trace of it was detected (Figure 1), as in the gastrointestinal tract, up to recent studies that report slight positivity even in locations so far considered ACE2 free, such as in circulating leukocytes (29)(30)(31). There is no question that the ACE2 receptor is also expressed at the level of epithelia of the respiratory system (tracheal and bronchial epithelial cells, alveolar epithelial cells, type 2 pneumocytes), cardiovascular system (endothelium of coronary arteries, myocites, epicardial adipocites, vascular endothelial, and smooth cells), gastrointestinal tract (esophagus keratinocytes, gastrointestinal epithelial cells, intestinal epithelial cells, duodenum, small intestine, rectum), urogenital system (kidney proximal tubules, bladder urothelial cells, luminal surface of tubular epithelial cells, testis, seminal vesicle), as well as in the liver and gallbladder and in the nervous system. (25,28,32) However, it is important to underline that, while mRNA seems to be expressed homogeneously in all tissues, the same is not always certainly for protein expression (Figure 1) (28). Many studies over the years have focused on the role of ACE2 in the cardiovascular system, both for the functions of the renin-angiotensin system (RAS) system and for the study of new therapeutic targets in cardiac pathologies (15,16). ACE2 is recognized as a protector of vascular tissues, balancing angiotensin II effects, protecting endothelia, and promoting mechanisms of regeneration (15). Intuitively, ACE2 impairments leads to severe cardiac dysfunction, with increased atherosclerosis, and endothelial damage. ACE2 is studied also in hypertension models, as genetic variation affects systolic function in men and ventricular mass in women (33). Not by chance, increased levels of ACE2, both at the gene level and protein expression, but also in circulating soluble forms, are detected after myocardial injury, suggesting a potential role as cardiac biomarker (15). Cardiac alterations result to be usually correlated with thyroid dysfunction, particularly to hyperthyroidism (34). Thyroid hormones also seem to act on ACE2 expression both influencing the receptor gene expression and conditioning the release of ACE from lung endothelium (35). ACE2 has been investigated also as cancer marker, as it has been observed an increase of ACE2 expression in thyroid cancer with an increase of ACE2/ACE ratio proportional to the differentiation grade of the cancer (36). The activity of ACE2 in cardiovascular system is strictly related to those in brain, as ACE2 is expressed in the neuronal area deputy to cardiovascular control, so that is result to be less expressed in case of cardiac injury, while an overexpression in brain leads to a protective action, via reduction of pro-inflammatory cytokines and augmentation of NO activity (37). Many animal models have been used for the study of ACE2 role in the brain. Data highlighted the antihypertensive and sympatholytic action of ACE2 in the hypothalamus via reduction in Ang II and increase in Ang-(1-7) levels, and a positive effect of ACE2 in the neuronal recovery from stroke (38). ACE2 is also involved in mechanism of memory, via regulation of brain-derived neurotrophic factor expression, and the production of reactive oxygen species, in stress response, regulating corticotropin releasing hormone at hypothalamus level, and in neurogenesis related to serotonin level, secondary to the availability of its precursor tryptophan (39,40). The fil rouge between tryptophan synthesis and ACE2 crosses the activity in many systems and binds their functionality. In fact, ACE2, involved in the RAS mediated homeostasis, plays at intestinal level regulating the microbiome, acting on amino acid uptakes, and expression of antimicrobial peptides (41). ACE2 acts as amino acid transporter, binding B0AT1, a neutral amino acid transporter, in the small intestine, and in animal model of ACE2 deficiency a reduction of tryptophan levels in the blood has been demonstrated (42). This reduction is reflected in the intestine with greater inflammation at the level of the colon, endothelium alteration and reduced ability to damage response, involving also mammalian target of rapamycin (mTOR) pathway, a member of the phosphatidylinositol 3-kinase-related kinase family of protein kinases (43). During the years attention was addressed also to the intestine, since it is known to express the highest level of ACE2 (28). In addition to ACE2 localization in the intestine, ACE2 was found in smooth muscle cells and endothelium of vessels from the stomach, and colon, smooth muscle cells of the muscularis mucosae, and the muscularis propria (28). ACE2 was also copiously present in the enterocytes of all parts of the small FIGURE 1 \| Schematic representation of ACE2 expression in human organs. ACE2 mRNA is present in all organs (28). ACE2 protein expression is present in heart, kidney, testis, lung (type I and type II alveolar epithelial cells), nasal, and oral mucosa and nasopharynx (basal layer of the non-keratinizing squamous epithelium), Frontiers in Medicine \| www.frontiersin.org 4 December 2020 \| Volume 7 \| Article 594495 FIGURE 1 \| smooth muscle cells and endothelium of vessels from stomach, small intestine and colon, in smooth muscle cells of the muscularis mucosae and the muscularis propria, in enterocytes of all parts of the small intestine including the duodenum, jejunum, and ileum (but colon), skin (basal cell layer of the epidermis extending to the basal cell layer of hair follicles smooth muscle cells surrounding the sebaceous glands, cells of the eccrine glands), endothelial, and smooth muscle cell of the brain (28). Red asterisk (*): ACE2 deficiency only hypothesized. intestine including the duodenum, jejunum, and ileum, but not in enterocytes of the colon (28). In addition to the gastrointestinal tract, ACE2 has been found in kidney and in pancreas, with dislocation similar to those of ACE2 that is in kidney apical surface area of the proximal tubules and pancreatic acini and islets (28). As for pancreas, the presence of ACE2 influences islets status via regulation of blood pressure and NO release, as well as acting on tissue fibrosis (44). The role of ACE2 has been widely investigated also for the onset of diabetes, as ACE2 deficiency has been associated with impairment of first-phase insulin secretion and of glucose tolerance (45)(46)(47). The alteration of RAS system and specifically of ACE2 activity induces an alteration of pancreatic islets, due to unbalance NO production, which in turn influences blood flow, also secondary to glucose availability (45)(46)(47). The wide expression of ACE2 in kidney is not so surprisingly, considering the pivotal role of RAS system in this organ, in which it regulates the electrolytic equilibrium via reabsorption of sodium and water into the blood, while causing excretion of potassium (48). ACE2 acts balancing the RAS activity, regulating renal homeostasis and it is postulated that its activity is more related to a local control than a systemic regulation of blood pressure (48). Effects of reduced ACE2 are described as promoting proteinuria, in particular albuminuria, glomerular disease and are related to diabetic nephropathy, with lower ACE2 expression at tubular level (48). Despite the role of ACE2 in hepatic glucose metabolism is not completely investigated, the alteration of the ACE2 pathway is, also in this localization, related to the development of impairment of metabolic activity, and in particular of insulin resistance (48). ACE2 in liver has been found expressed in endothelial cells, bile duct cells, and perinuclear hepatocytes and it was mostly elevated in hepatic fibrogenic resistance (28). Notably, insulin resistance correlates with endothelium-dependent and insulinmediated vasodilatation (46,49). In addition, a recent RNAseq data in the human protein atlas database have shown the highest expression of ACE2 in liver cholangiocytes, followed by hepatocytes (50) ACE2 expression seems to be correlated to the sensory organs. However, the real expression of ACE2 at ocular level, instead, seems to be still object of debate. Although it is the least widely expressed among the RAS system components, ACE2 is detectable in the aqueous humor (51,52). Some papers declare a not significant mRNA presence and immunoreactivity of ACE2 in human conjunctiva (53), while according to others, ACE2 gene expression is detectable both in human conjunctiva and primary pterygium tissues, even if in a reduced cohort of patient (54). ACE2 is expressed at the oral level in particular at the oral tongue than at the buccal and gingival and could be found in epithelial cells, T cells (<0.5%), B cells (<0.5%), and fibroblast (<0.5%) (31). In addition, in the oral and nasal mucosa and in the nasopharynx, ACE2 expression was found in the basal layer of the non-keratinizing squamous epithelium (55). Human ACE2 was detected in ciliated airway epithelial cells of human airway tissues derived from nasal regions (55). Concerning ACE2 presence at the ear level no data were present on human. However, a recent online study found high expression of ACE2 in the cat ear tip (56). Another sensory organ where ACE2 was also found is at the skin level (57). The activity of RAS system in controlling cell proliferation and differentiation, also in case of tissues injury in the mechanism of self-renewed of damaged cells and neoangiogenesis, is reflected also in the skin, where the epidermal stem cells express the different players of this system, including ACE2 (57). Immunohistochemical evaluation of ACE2 presence in healthy and oncologic patients showed ACE2 in basal cell layer of normal epidermis and sebaceous glands and a reduction of ACE2 reactivity in patients affected by pre-malignant lesions (actinic keratosis) and non-melanoma malignant skin cancers (basal cell carcinoma and squamous cell carcinoma), suggesting an involvement in the pathogenesis of the disease (58). Considering the role of Angiotensin II in the menstrual cycle, the presence of ACE2 in the female reproductive systems appear quite intuitive. In fact, AngII acts on follicular, ovulatory and luteinic phases, influencing follicle development, oocytes maturation, and corpus luteum progression, balancing the levels of steroid hormones (59). In addition, it promotes spiral artery vasoconstriction and endometrium regeneration at the uterus level. Angiotensin II has been identified also as a player in endometrium fibrosis and endometrial metastases (59). Not surprisingly, during pregnancy, ACE2/AngII/Ang 1-7 axis is involved in maintenance of blood pressure and alterations of this pathway correlate with disorders like pre-eclampsia and eclampsia, while reduction of ACE2 expression negatively influences gestation and fetus birth (60). In parallel, ACE2 expression has been detected also in testis, particularly in spermatogonium and Leydig and Sertoli cells, with possible correlation with spermatogenesis and maintenance of functional and structural integrity of the apparatus (61). Finally, despite the presence of ACE2 in numerous organs, tissues and cells have not been completely clarified and in many of them not yet investigated, ACE2 seems to be absent in the spleen, thymus, lymph nodes, bone marrow, and in several cells of the immune system (15,62). However, it is important to underlined that numerous studies on ACE2 expression in bone marrow are currently in progress since all the players of RAS system are present in the bone marrow, acting on cell lineages proliferation and also in hematopoietic restoration after myelosuppression, and ACE2 seems to have in particular a role in CD34+ proliferation (63). In this moment, with the ongoing COVID-19 pandemic, this rapid overview related to the distribution, expression and activities of the ACE2 in human body could help and improve our understanding on potential infection routes of SARS-CoV-2 through the body. Thus, in the next section we discuss how the presence, distribution and abundance of ACE2 in specific target organs may be related to the COVID-19 clinical symptoms and manifestations. Nasal Cavity On October 5, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND (Nose OR Nasal Cavity)" we found 388 papers. Most of the studies were guidelines on how to perform nasal and oropharyngeal swab procedure for the screening of COVID-19 infection. The other studies detected, analyzed and discussed the different nasal manifestations in COVID-19 patients (64). Orhinolaryngological symptoms resulted common manifestations of COVID-19, particularly in mild or moderate form of the disease (65,66). The nasal cavity and turbinates have important physiological functions in filtering, warming, and humidifying inhaled air and these functions are critical during SARS-CoV-2 infection since the nasal cavity is the principal gateway for virus entrance. In fact, epithelial cells in this region are considered suitable clinical sample for early virus detection. Increasing number of reports on SARS-CoV-2 positive patients described olfactory dysfunction, such as loss of smell, cacosmia, phantosmia nasal obstruction or rhinorrhea, and nasal congestion (1, 67-73) 1 . Some data also reported SARS-CoV-2 positive patients with isolated anosmia, without any other symptoms, suggesting these patients as a potential source of rapid virus spread (68,69,74). Anosmia in SARS-CoV-2 positive patients can be present as primary symptom or as an early symptom, with different percentage among the examined studies, percentages that can range from 6 to ∼80% (1, 70-73, 75). Lechien et al. demonstrates that about 87% patients with an anosmia duration ≤12 days were also PCR SARS-CoV-2 positive (73). Additionally, Kaye et al. analyzing a court of 237 COVID-19 patients showed that 73% of patients reported anosmia and 26.6% reported loss of sense of smell as initial symptom (70). Patients below 40 years, particularly female, seem to be the more prone to develop SARS-CoV-2 form with only hyposmia/anosmia manifestations (1,(69)(70)(71)(72). However, an Asian study reported a lower percentage of patients with olfactory dysfunctions in comparison to European patients (73). This aspect may be probably due to the diverse ACE2 polymorphisms and expression level between Asian and European individuals (74). The loss of smell in SARS-CoV-2 patients may be caused by different factors, such as localized olfactory cleft oedema, architectural deformity of the olfactory neuroepithelium, or direct neuro-invasion of the olfactory nerve pathways. As above described, it is important to underline that gene expression databases highlighted a moderately/high expression of ACE2 in human olfactory mucosa (76). However, to date, whether ACE2 expression in the olfactory epithelium is neuronal or non-neuronal or if it occurs in both cell types it is not completely clear (77,78). SARS-CoV-2 brain infection could be facilitated by the neuronal expression of the host receptors through absorption in ciliated dendrites/soma and consequent axonal anterograde transport along the olfactory nerve (79,80). Concerning the non-neuronal expression of ACE2, it could be due to the nasal cavity olfactory epithelium that would work as virus reservoir (79,80). Several RNaseq transcriptome reports conducted in human and murine olfactory epithelium suggested a non-neuronal expression of ACE2 as well as of TMPRSS2 (79)(80)(81)(82), but further studies are mandatory to confirm these finding. It was also shown that nasal brushing epithelial cells, nasal turbinate epithelial cells, and nasal airway epithelial cells contained ACE2expressed and TMPRSS2-expressed cell clusters (82). Oral Cavity On October 5, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND oral cavity" we found 218 papers. Several studies evaluated the presence of SARS-CoV-2 in saliva through entry into the oral cavity with several potential pathways, via a direct infection of oral mucosa lining cells, via droplets from the respiratory tract, from the blood circulation by gingival crevicular fluid, or via extracellular vesicles released from infected cells and tissues (83). To et al. confirmed that SARS-CoV-2 can be detected by PCR in about 92% of saliva samples, indicating the saliva as a potential source of SARS-CoV-2 spreading (84). The presence of SARS-CoV-2 in patients' saliva also suggested the likelihood of salivary gland infection. Chen et al. collecting saliva directly from salivary gland, found SARS-CoV-2 nucleic acid, hypothesizing that salivary glands were SARS-CoV-2 infected (85). This hypothesis is further reinforced by the fact that the ACE2 epithelial cells of the salivary glands have been shown to be an early target for the SARS-CoV-2 (86). In addition, high levels of mRNA and protein levels of the cellular protease Furin as well as of TMPRSS2 have also been found in the salivary glands (86). Thus, the possible role and function of salivary gland cells in the initial SARS-CoV-2 entry and progress must be further considered and validated as well as their potential function as virus reservoir, able to establish a persistent infection which could last also for months (87). Furthermore, it should be underlined that the saliva samples not only contain saliva secreted from the salivary glands but also the secretions from the nasopharynx and from the lung via the action of cilia lining the airway. Thus, more studies are needed to delineate the real sources and functions of SARS-CoV-2 in saliva. Another point related to the oral cavity is represented by the fact that numerous studies reported an acute loss of taste (hypogeusia/ageusia) as a frequent symptom of SARS-CoV-2 infection, particularly common among females and younger individuals (∼20-39 years) (1,84,85). A recent case series presented several cases of SARS-Cov-2 infection where the loss of taste was also associated with oral lesions (88). These lesions presented two distinct patterns, one resembling aphthouslike ulcers in young patients with mild cases of COVID-19 and another with more widespread patterns resembling Herpes Simplex Virus 1 necrotic ulcers in the more severe and immunosuppressed older individuals (89). Whether these lesions were due directly by SARS-CoV-2 or were an associated manifestation resulting from the severe compromised state of the patient remains to be determined. However, what is known is that taste disorders are linked to an extensive variety of viral infections (90). Upper respiratory tract infection can lead to acute onset ageusia because of viral damage to the olfactory epithelium (90). Furthermore, as previously reported for the nasal cavity, viruses can also use the olfactory nerve as a route into the central nervous system (CNS). Thus, ageusia may be a secondary result of olfactory dysfunction. However, it is important to underline that ACE2 is not only extensively express in the salivary glands and in oral tissues and its expression was higher in tongue than buccal or gingival tissues (https://gtexportal.org). Furthermore, ACE2 positive cells were enriched in epithelial cells, thus damage of mucosal epithelial cells of the oral cavity may explain ageusia, oral mucosal ulcerations, and necrosis detected in SARS-CoV-2 patients (31,91). In addition, it was reported that the ACE2 within oral mucosa is also expressed in lymphocytes, and comparable results were also reported for other organs of the digestive system (31). However, since the ACE2-positive lymphocytes is quite few whether this aspect could indicate that SARS-CoV-2 attacks the lymphocytes and leads to the severe disease in patients' needs further studies (31). More in generally SARS-CoV-2-mediated gustatory disturbances has not yet been definitively identified. Eyes On October 5, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND (eyes OR ocular manifestations)" we found 820 papers. Most of the studies were official recommendations of ophthalmological societies for precaution and prevention of SARS-CoV-2 infection or studies on the impact of COVID-19 outbreak on eye care. Currently, the presence and prevalence of ocular manifestations in SARS-CoV-2 infection, consistent with conjunctivitis and including conjunctival hyperemia, chemosis, epiphora, or increased secretions, are still controversial (92)(93)(94)(95). Despite it was reported that only a small percentage (from about 1 to 6%) of SARS-CoV-2 positive patients developed signs of conjunctivitis, other studies showed that up to 31% of SARS-CoV-2 hospitalized patients presented conjunctivitis (96)(97)(98)(99). Wu et al. showed that about 31.6% of COVID-19 patients had ocular abnormalities, and ocular symptoms were more frequent in severe cases of COVID-19 patients (97). In fact, about 50% of COVID-19 patients with ocular abnormalities were classified as critical, 16.7% were classified as severe, and 33.3% were classified as moderate severity (97). In addition to conjunctivitis other ocular abnormalities directly correlated with the COVID-19 severity seem to be alterations in retina and in its vasculature (100). A recent study evaluating the retina of patients with COVID-19, within 30 days from the onset of systemic symptoms, found an enlargement of retinal arteries and veins in more severe cases and showed an inverse correlation with time to symptoms onset (100). In this context, Casagrande et al. demonstrate the existence of SARS-CoV-2 nucleic acid in the human retina COVID-19 patients (101). Additional studies also highlighted the presence of SARS-CoV-2 RNA in tear film and/or conjunctival swabs of COVID-19 patients with conjunctivitis but not in patients without ocular symptoms (98,(102)(103)(104)(105). Differently, Xie et al. demonstrated that the SARS-CoV-2 RNA was present also in the normal ocular surface of COVID-19 patients without conjunctivitis (106). Despite this point is still debated, it is critical to underline that ocular surfaces have a great tropism for respiratory viruses and also for coronavirus (107,108). Whether specifically SARS-CoV-2 may infect retina and conjunctival cells in human remains unclear. Based on the current literature, several reports hypothesized that the exposure of the ocular surface to SARS-CoV-2 could lead to infection probably due to the drainage of virus particles via the nasolacrimal duct, specifically through the lacrimal canaliculi that drain tears from the eye surface into the nasal cavity, into the respiratory tract (109,110). In this context it is important to emphasized that others reports also considered the presence of ACE2 and TMPRSS2 on the cornea and conjunctiva as a possible virus route (56,111,112). The presence of the ACE2 and TMPRSS2 on the corneal cells may allow the virus to cross the ocular surface, and then spread from the eye to other parts of the body through the blood stream and/or through the nervous system (ophthalmic branch of trigeminal nerve) (112,113). However, to date, there are no clear evidences that SARS-CoV-2 virus, in humans, can enter inside the eye or spread to the brain through corneal nerves (114). Lungs On October 5, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND lungs" we found 4,138 papers. While SARS-CoV-2 was detected in many organ systems, the lungs seems to be the main organs affected by the virus (96,115). In fact, it is known that the upper respiratory tract and lungs serve as predominant site of virus entry and replication and that SARS-CoV-2 patients showed the symptoms of pneumonia and alveolar damage (116). The most common and severe complication in patients with SARS-CoV-2 patients is acute hypoxemic respiratory failure or acute respiratory distress syndrome that lead to oxygen and ventilation therapies (1,(117)(118)(119)(120)(121)(122)(123)(124)(125)(126)(127)(128)(129)(130)(131)(132)(133)(134)(135). Some of these critically ill patients also required intubation and invasive ventilation. Lungs radiological images and computed tomography (CT) scans of SARS-CoV-2 positive patients provided numerous information about the severity of the infection and showed abnormal results in about 86% of patients (1,(117)(118)(119)(120)(121)(122)(123)(124)(125)(126)(127)(128)(129)(130)(131)(132)(133)(134)(135). The most common patterns of radiological images and CT scans were groundglass opacities, consolidation, centrilobular nodules, architectural distortion, bronchial wall thickening, vascular enlargement, traction bronchiectasis, reticulation, crazy paving pattern, intrathoracic lymph node enlargement, and subpleural bands, that cause pulmonary discomfort and require rapid diagnosis and treatment (1,(117)(118)(119)(120)(121)(122)(123)(124)(125)(126)(127)(128)(129)(130)(131)(132)(133)(134)(135). In addition, autoptic results revealed that in about 48% of cases the predominant histopathological finding were capillary congestion, microthrombi as well as moderate intra-alveolar fibrin exudation resultant in exudative disseminate alveolar damage and superimposed bronchopneumonia (135). A more widespread histological pattern of alveolar damage with greater fibrotic evolution in the lungs was observed in patients who died after a long period of mechanical ventilation (136). In few cases, an intra-alveolar deposition of neutrophilic granulocytes, probably due to superimposed bacterial infection, was also detected (137). Since the distribution of ACE2 in different organs seems to be notably linked to the clinical symptoms of SARS-CoV-2 infection and since the acute respiratory distress syndrome is a potential deadly complication of SARS-CoV-2, research studying lung complications of ACE2 down-regulation are of key significance in this context. Several studies on lung injury highlighted that ACE2 receptors downregulation lead to critical inflammatory lesions in the respiratory tract (alveolar wall thickening, edema, infiltrates of inflammatory cells, bleeding) which seem to be carried out by angiotensin II (135,(138)(139)(140)(141)(142). A key point to remark is that the wide surface of alveolar epithelial cells might explain the vulnerability of lungs to the virus invasion. As previously explained ACE2 are principally expressed in type II pneumocytes, little cylindrical cells that correspond to the 5% of all pneumocytes (1). These type of pneumocytes exert immunoregulatory functions and are of key importance for alveolar surfactant production and they also function as stem cells, progenitors of type I pneumocytes, that represent the 95% of all pneumocytes and that are responsible of gas exchanges (142). Thus, the damage of type II pneumocytes owing to the binding of SARS-CoV-2 to ACE2 receptors is critical for several factors, i.e., for the local unopposed ACE→ Angiotensin II→ AT1 receptor axis over-activity, for the reduced production of alveolar surfactant by injured type II pneumocytes that lead to reduced lung elasticity and, finally, for the reduced repair of type I pneumocytes that bring to impaired gas exchanges and fibrosis (143). While ACE2 is expressed in the bronchial epithelium and in type 2 pneumocytes, TMPRSS2 results strongly expressed in the cytoplasm of bronchioles and alveolar epithelial cells (144). Since ACE2 was found to exist on alveolar epithelial cells at approximately similar level as in the whole lung, Sato et al. found that the expression level of TMPRSS2 was considerably different between the peripheral and central parts of the lung (145). Thus, since that peripheral parts of the lung strongly express TMPRSS2, along with ACE2, the SARS-CoV-2 may be considered to damage the peripheral area at the beginning of infection. These data explain why chest CT revealed consolidation and ground glass opacities in the bilateral peripheral lobes in COVID-19 cases (146). However, these factors would not even prevent the simultaneous role of other mechanisms including an altered immune response to initial viral invasion, or a genetic susceptibility to hyper-inflammation and thrombosis (8,147). In SARS-CoV-2 pneumonia, thrombosis may play a direct, and critical role in gas exchange abnormalities and in multisystem organ dysfunction. Unfortunately, to date, as for all the other organs affected by SARS-CoV-2, the lungs impairment during this new infection remain to be further clarified. Heart and Blood Vessels On October 5, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND (cardiovascular system OR heart OR blood vessels)" we found 3,170 papers. In most of these reports cardiovascular complications emerged among the most significant manifestations in SARS-CoV-2 infection (148)(149)(150)(151)(152)(153)(154)(155). Different cardiovascular complications, such as myocarditis, acute coronary syndrome, decompensated heart failure, pulmonary embolism, cardiogenic shock, and infection of a heart transplant recipient, accompanied by altered levels of creatine kinase isoenzyme-MB, myohemoglobin, cardiac troponin I, and N-terminal pro-brain natriuretic peptide were currently reported (1,(149)(150)(151)(152)(153)(154)(155). In addition, a high prevalence of pre-existing cardiovascular morbidities, including hypertension, and coronary artery diseases, has been detected among patients with severe SARS-CoV-2 (1,149,156). In COVID-19 patients, the highest mortality rates were also observed in case of pre-existing cardiovascular disease and elevated cardiac troponin levels (137,157). Furthermore, patients with higher troponin levels had also increased markers of inflammation, including C-reactive protein, interleukin (IL)-6, ferritin, lactate dehydrogenase (LDH), high neutrophil count, and high amino-terminal pro-B-type natriuretic peptide (158). Despite it was initially hypothesized that COVID-19 patients with pre-existing cardiovascular morbidities and treated with ACE inhibitors (ACEi) or angiotensin receptor blockers (ARBs) (155,159,160). could be at increased risk for severe SARS-CoV-2 infection, a recent retrospective study on COVID-19 patients with hypertension showed that ACEI/ARB therapy attenuated the inflammatory response (161). In addition, a study on SARS-CoV-2 patients with hypertension showed no difference in the percentage of patients treated with ACEi/ARBs between those with severe and non-severe infection and between survivors and non-survivors (162). However, understanding the positive or negative effect of ACEi/ARB in COVID-19 appears to be complex, and this could also be due to the clinical stage of the virus (viral contamination phase vs. tissue inflammation phase). Several clinical trials on this question are forthcoming (NCT04329195, NCT04331574, NCT04351581, NCT04353596). To date, the mechanisms by which SARS-CoV-2 leads to cardiac manifestations is currently unclear. These mechanisms would involve several factors such as a direct viral damage and an immune-mediated damage by inflammatory cytokines (i.e., a systemic cardiotoxic cytokine-storm), and cytotoxic immune cell response. As reported in the previous section, the cardiac tissue presents a high ACE2 expression level (163). Specifically, it was shown that cardiomyocytes from the heart contain about 6% ACE2-expressed cells and 0.8% TMPRSS2-expressed cells, and the cardiovascular progenitor cells contain 12.5% ACE2-expressed cells and 0.4% TMPRSS2-expressed cells, thus SARS-CoV-2 could directly infect the myocardial tissue (82). In addition, Furin can also be considered a critical molecule that makes SARS-CoV-2 cause adverse cardiovascular events through the ACE2 receptor. This speculation is supported by the occurrence of high level of Furin in the peripheral blood of COVID-19 patients (164). Additionally, PCR analyses also identified SARS-CoV-2 in the cardiac tissue of ∼35% of infected patients, further supporting that a direct viral damage can occur (165). Kuba et al. by using a mouse model showed that SARS-CoV pulmonary infection leads to an ACE2-dependent myocardial infection (138). This infection can lead to a localized inflammatory response with resulting myocarditis that bring to acute cardiac injury and the prospective for arrythmias or heart failure (166). Autoptic data on SARS-CoV-2 positive patients showed the existence of mononuclear inflammatory myocardial infiltrate, thus supporting this hypothesis also for this new coronavirus (3). Numerous studies also reported immunological derangements in SARS-CoV-2 positive patients (167,168). This altered immunologic status has been related with an increased risk of cardiovascular disease and could be also an indirect mechanism of immunological dysfunction that lead to cardiac sequelae (166)(167)(168). In addition, numerous SARS-CoV-2 positive patients showed respiratory distress that lead to hypoxemia that could cause cardiac injury secondary to an oxygen mismatch (166)(167)(168). Other systemic consequences of cardiac injury in SARS-CoV-2 patients could also be related to sepsis and disseminated intravascular coagulation (DIC) that vary from minimal change to interstitial inflammatory infiltration and myocyte necrosis vasculature microthrombosis and vascular inflammation (166)(167)(168). However, to date whether SARS-CoV-2 infection impair the heart remains to be further demonstrated. Kidney and Bladder On October 6, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND (kidney OR urinary system)" we found 1,031 papers. The kidney is one of the major organs which play a key role in the filters which excrete toxins, waste products, and extra water from our body. Despite most of the work were focused on kidney transplantation and on the management of dialysis patients during SARS-CoV-2 infection, several studies reported an increased incidence of acute renal injury during the infection (169,170). The bladder may also be affected and may ultimately lead to multiple-organ failure and death (169,170). Although initial reports suggested that the burden of acute kidney injury during SARS-CoV-2 infection was moderately low (about 0.5%), recent studies reported an incidence going up to 56.9% (115,169,(171)(172)(173)(174). In critically ill patients, this incidence was remarkably higher, ranging from 61 to 76% (175). A higher incidence of acute renal injury has been reported in USA and UK than in China (96,150,174,176). Several studies also showed that patients with acute renal injury have a higher mortality rate compared to other patients and this is particularly true for those in the ICU (177)(178)(179). In a recent study, it was shown a high incidence of renal dysfunction (46%) and acute renal injury (29%) also in hospitalized children with COVID-19 (180). Patients with acute renal injury also showed elevated levels of serum creatinine and blood urea nitrogen associated to higher leukocyte count and lower lymphocyte and platelet counts (169). Prolonged activated partial thromboplastin time and higher D-dimer, both coagulation parameters, were also more common in these patients (169). In addition, a high percentage of SARS-CoV-2 patients with acute renal injury had proteinuria albuminuria and hematuria, along with isolation of viral RNA from urine, all factors that support the potential viral tropism for the kidney (181,182). This tropism was also confirmed from an autopsy study by Su et al. that demonstrated by electron microscopy SARS-CoV-2 presence in the renal tubular epithelium of seven of 26 SARS-CoV-2 patients (176). This study also showed the presence of a diffuse proximal tubule injury with the loss of brush border, non-isometric vacuolar degeneration, necrosis, and occasionally hemosiderin granules and pigmented casts (176). In addition, a prominent erythrocyte aggregates obstructing the lumen of capillaries without platelet or fibrinoid material were also detected (176). Clusters of coronavirus-like particles with distinctive spikes in the tubular epithelium and podocytes were also detected. Post-mortem examination of the viral nucleocapsid protein in situ in the kidney also showed that SARS-CoV-2 antigens is accumulated in kidney tubules, suggesting that SARS-CoV-2 may infects kidney directly, leading to acute renal injury and potentially contributing to viral spread (183)(184)(185). This direct route of SARS-CoV-2 may be due to an ACE2-dependent pathway. It was found that both proximal tubular cells or tubular progenitor cells in the kidney co-expressed ACE2 and TMPRSS2 and their expression levels resulted high in nephron epithelial cells, epithelial cells, endothelial cells, and mesangial cells of the kidney (82,186). Additionally, Pan et al. showed that the TMPRSS2 gene was co-expressed with ACE2 in kidney podocytes (170). These cells are particularly vulnerable to viral infection and their injury easily induces heavy proteinuria that was detected in about 43.9% of SARS-CoV-2-infected patients (181). The coexpression of ACE2 and TMPRSS2 in renal tubular cells could imply that SARS-CoV-2 may directly bind to ACE2-positive cells in the kidney and destroy the function of renal tubules. However, kidney disease involvement in SARS-CoV-2 patients is likely to be multifactorial and may be also due to cytokine damage (high levels of IL-6), organ crosstalk (Lung-kidney) and other systemic effects (187,188). Stomach and Intestines On October 6, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND (stomach OR intestines OR gastrointestinal system OR digestive system)" we found 977 papers. A lot of studies showed that the gastrointestinal tract represents a common target organ of SARS-CoV-2 infection (1,2,29,96,(150)(151)(152)(189)(190)(191)(192)(193)(194)(195)(196)(197)(198). A recent study suggests that the gastrointestinal symptoms in COVID-19 patients can be present up to 50% (39.6-50%), with symptoms including nausea, diarrhea, anorexia, abdominal pain, belching, and emesis (199,200). Anorexia appears to be the most common gastrointestinal symptom (26.8%), but the mechanism of its onset in COVID-19 patients remains unclear (4). However, this symptom can be due to gustatory dysfunction, which was found in a high percentage of COVID-19 patients (201). Several data reported that these gastrointestinal manifestations during SARS-CoV-2 infection can be associated with a poor disease course; comparing patients with non-severe disease with those with severe infection it was shown a higher risk of developing gastrointestinal symptoms in patient with severe infection (29,202). The occurrence of these gastrointestinal symptoms can not only coexist with other symptoms, but also precedes the typical phenotype of SARS-CoV-2 infection (203). It was shown that also pediatric patients and children with SARS-CoV-2 infection may present digestive symptoms, most commonly diarrhea, in the absence of respiratory symptomatology (203,204). Although different clinical features, such as milder disease course symptoms are present in pediatric patients and children with SARS-CoV-2, the gastrointestinal symptoms appear to be similar to those found in adult individuals (204). Despite gastrointestinal symptoms were frequently observed in SARS-CoV-2 patients, to date, the exact significance of these manifestations are still unclear. An autopsy report, with details of gastrointestinal pathology in a SARS-CoV-2 patient, showed the presence of segmental dilatation and stenosis in the small intestine (205). To date, autopsy data and reports with a full description of the gastrointestinal appearance associated to SARS-CoV-2 infection are still few to allow a clear conclusion. In addition to the clinical symptoms induced by the gastrointestinal disorders during SARS-CoV-2 infection, these manifestations can highlight one more route of virus transmission, i.e., the fecal-oral transmission. An increasing number of data showed that stool samples contain high concentration of SARS-CoV-2 RNA during infection for a relatively long period of time (from 1 to 12 days) (193,204,206). These data were also confirmed in pediatric patients and in children where ∼80% of patients resulted positive on rectal swabs even after negative nasopharyngeal tests (204). This aspect suggests a potential replication of SARS-CoV-2 virus in the gastrointestinal tract. This hypothesis is partially confirmed by Lin et al. that analyzing by endoscopy severe and nonsevere SARS-CoV-2 patients with gastrointestinal manifestations detected the presence of SARS-CoV-2 RNA in esophagus, stomach, duodenum, and rectum of severe patients while only in the duodenum on one of four non-severe patients (4). Although, there are numerous data on gastrointestinal symptoms during SARS-CoV-2, the exact mechanism by which the virus affects the gastrointestinal tract is still not so clear. The occurrence of several mechanisms has been hypothesized. One mechanism may involve the presence of ACE2 receptors in the gastrointestinal tract. Liang et al. found that ACE2 was highly expressed in the small intestine especially in proximal and distal enterocytes (207). In addition, Zhang et al. found that ACE2, TMPRSS2, and Furin, all critical for fusion of viral and the cellular membranes, were co-expressed in esophageal upper epithelial and gland cells and also in the enterocytes from ileum and colon, thus speculating exactly these organs as potential targets for SARS-CoV-2 (208). In addition, Guo et al. suggested that TMPRSS2 was highly expressed in almost all organs of the digestive tract including colon, stomach, small intestine, and esophagus (209). The coexpression of ACE2 and TMPRSS2 in the intestinal enterocytes may explain the disruption of intestinal absorption that leads to diarrhea. However, a second mechanism could involve a direct injury of the gastrointestinal system due to an inflammatory response (cytokine storm) (208). Absorptive enterocytes may be infected and destroyed by the virus, probably leading to malabsorption, disturbed intestinal secretion, and an activated enteric nervous system ensuing symptoms like diarrhea (210). Liver On October 6, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND liver" we found 1,319 papers. Several data reported that approximately half of SARS-CoV-2 patients show liver biochemistry abnormalities, with increased levels of aminotransferases, gamma-glutamyl transferase, bilirubin, and alkaline phosphatase (116,(211)(212)(213)(214)(215)(216)(217)(218). Median aspartate aminotransferase-dominant aminotransferase increase seems to indicate the disease severity and seems to be an index of hepatic injury (211). Concerning hepatic injury, Bloom et al. reported that about 1 in 5 patients developed grade 3 or 4 hepatocellular injury during hospitalization (212). In addition, it was reported that liver abnormalities seem to be more common in patients with severe disease upon presentation (212). In fact, a recent meta-analysis including 20 retrospective studies with 3,428 COVID-19 patients revealed that higher levels of alanine aminotransferase, aspartate aminotransferase and bilirubin were associated with a significant increase in the severity of COVID-19 infection (219). A recent meta-analysis also linked elevated admission levels of these markers to patient mortality (220). Other common factors linked with liver injury were decreased lymphocyte count, increase neutrophil count, and male gender (213). However, to date, the exact changes that lead to the altered liver biochemistries in SARS-CoV-2 patients remains unclear. Post-mortem liver biopsy showed the presence of a moderate microvascular steatosis and a mild lobular and portal activity (116). Another study suggested collateral liver damage from viral-induced cytotoxic T-cells (221). Additionally, since also abnormal coagulation markers have been reported in SARS-CoV-2 patients it is possible that the presence of microthrombi lead to an altered hepatic perfusion and consequent hepatocyte injury and aspartate aminotransferase increase (214,215,222). Whether these changes can be due to direct viral cytopathic effect, to cytokine release linked with SARS-CoV-2, to ischemia, to a preexisting condition, or to other causes, such as drug-induced liver injury, are currently unknown, also because studies on mechanisms of SARS-CoV-2 related liver dysfunction are limited. What we know currently is that ACE2 receptor are highly expressed in cholangiocytes (59.7%) and low expressed in hepatocytes (2.6%), thus some studies hypothesized a cholangiocytes mediating viral-associated injury (216). However, Zhou et al. showed that TMPRSS2 is highly expressed in hepatocytes (223). In fact, it was shown that alkaline phosphatase, an index of cholangiocytes injury, was the liver parameter less subject to significant alterations during SARS-CoV-2 infection while, aminotransferases and gammaglutamyl transferase, indicators of hepatocyte injury, were the more common and almost always altered liver parameters in severe SARS-CoV-2 patients (116). In fact, autoptic analyses of liver tissue from SARS-CoV-2 patients do not demonstrate a cholangiocyte damage (116). As just described, liver injury in COVID-19 may be the direct insult to the liver or bile cells via receptors of ACE2 but it is further aided by hyper-inflammation, cytokine storm or bystander hepatitis and drug-induced damage. Another hypothesis is that since the SARS-CoV-2 RNA was also present in stool, it would be possible a transmission from the gut to liver by portal circulation (224). To date the exact mechanism of viral-associated liver injury needs further investigation. Gallbladder On October 7, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND gallbladder" we found 21 papers. Despite few articles were found on gallbladder during SARS-CoV-2 infection, several information on its alteration during the new viremia were found in manuscripts on liver injury (225,226). Gallbladder is a storage pouch for bile that is continually produced by liver, thus their functions are strictly related. Specific right upper quadrant ultrasounds on gallbladder of SARS-CoV-2 patients detected gallbladder sludge and distention in about 54% of patients, suggesting the presence of cholestasis (226,227). Cholestasis in these patients seem to be not associated with age, gender, ICU admission, or gastrointestinal symptoms at presentation (226). The fatality rate seems to be higher among patients with cholestasis than those without cholestasis (228). As for liver, the gallbladder was found susceptible to the infection probably due to the high ratio of gallbladder epithelium cells expressing ACE2 (28). Also in this case the mechanism of viral-associated gallbladder alterations is unclear, although it is obvious that its alterations during SARS-CoV-2 infection are associated with liver injury. Pancreas On October 7, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND pancreas" we found 77 papers. Currently, data on pancreas involvement in SARS-CoV-2 infection are scarce. However, several case reports showed pancreatic injury in COVID-19 patients and it was reported that about 1-2% of nonsevere and 17% of severe patients with SARS-CoV-2 infection presented pancreatic injury (5,(229)(230)(231). Several of these patients also presented abnormal blood glucose, suggesting that the pancreatic injury might be due directly to cytopathic effect by local SARS-CoV-2 replication (5,(229)(230)(231). Additionally, pancreatic injury might be caused indirectly by systemic responses to respiratory failure or to the harmful immune response induced by SARS-CoV-2, which led also to the damage in multiple organs (5). Similar results were also found by Liu et al. that detected elevated levels of amylase and lipase associated to focal enlargement of the pancreas or dilatation of the pancreatic duct based on CT scans (232). Hadi et al. also described SARS-CoV-2 patients with severe acute pancreatitis, which itself may lead to multi-organ failure including adult respiratory distress and kidney failure as seen the patients examined in the study (231). Considering the proportion of SARS-CoV-2 patients with pancreatic injury and theexpression of ACE2 and TMPRSS2 in the pancreas (particularly in pancreatic islet cells), researcher and clinicians should pay attention to the possibility of damage caused by SARS-CoV-2. Brain On October 7, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND brain" we found 1,293 papers and most of them showed that SARS-CoV-2 invades the CNS, developing neurological impairments such as stroke, epilepsy, anosmia and hypogeusia, seizures, and encephalitis (1,66,(233)(234)(235)(236). Specifically, a retrospective analysis by Mao et al. (237) underlined that about 40% of SARS-CoV-2 patients developed headache, disturbed consciousness, and other brain dysfunction symptoms (1), and an autopsy study reported the presence of edema in brain tissue of SARS-CoV-2 patients (66). Several case-series and two retrospective studies also reported critical stroke conditions related to COVID-19 (238,239). In this context, Beyrouti et al. examining a small cohort of COVID-19 patients, also underlined that ischemic stroke (confirmed by reverse-transcriptase PCR) linked to severe SARS-CoV-2 patients occurs in the context of a systemic highly prothrombotic state, as shown by large vessel occlusion and elevated D-dimer levels (240), conditions that can make patients more prone to acute cerebrovascular events. Moriguchi et al. reported the first case of meningitis related to COVID-19 underling that SARS-CoV-2 RNA was not present in nasopharyngeal swab, but it was detected in cerebrospinal fluid sample (241). Numerous cases of encephalitis/encephalopathy associated with SARS-CoV-2 infection were also described and were confirmed by postmortem analyses, where acute disseminated encephalomyelitis and neocortical micro-infarcts were detected (242)(243)(244). Neurologic complications associated to COVID-19 patients are not limited to the CNS. In fact, several authors also reported a correlation between SARS-CoV-2 and Guillan-Barrè syndrome, an acute/sub-acute immune-mediated polyradiculoneuropathy with diverse degrees of limbs or cranial-nerves weakness, lack of deep tendon reflexes, sensory, and dysautonomic symptoms cause by peripheral nerves and roots demyelination and/or axonal injure (245)(246)(247)(248). Other studies also described Miller Fisher syndrome as another neurologic complication of SARS-CoV-2 infection (249)(250)(251). These neurological manifestations in the brain of SARS-CoV-2 infected patients were confirmed and recognized by CT scan images and magnetic resonance imaging (MRI) scan, where presence of necrotizing hemorrhagic encephalopathy, brain thrombosis and acute infarction, eptomeningeal enhancement, perfusion abnormalities, and cerebral ischemic stroke, demyelinating lesions, right temporal lobe edema, and brainstem inflammation, were recognized (252)(253)(254)(255). In addition, the presence of SARS-CoV2 was identified in frontal lobe tissue by using transmission electron microscopy (237) and by genome sequencing in cerebrospinal fluid of SARS-CoV-2 patients, supporting that this new pneumonia virus can cause nervous system damage (241). In addition to the above described neurological manifestations, several SARS-CoV-2 infected patients showed delirium and/or mental status changes. These symptoms may be a manifestation of direct CNS invasion, induction of CNS inflammatory mediators but may be also a secondary effect of other organ system failure, an effect of sedative strategies, a prolonged mechanical ventilation time, or environmental factors, including social isolation (256). Despite these data clearly highlighted the involvement of the brain in SARS-CoV-2 infection, the exact mechanism for virus neurotoxicity is not yet straightforward, since this depend on the brain entry route of the virus, which, to date, has not been fully elucidated (257). The pathway of the virus into the brain could be primarily linked to the route of transmission and distribution of intracellular receptors of SARS-CoV-2. Mao et al. hypothesized that SARS-CoV-2 virus may interact with ACE2 in the capillary endothelium and caused blood-brain-barrier destruction, thus promoting the entry of the virus into CNS (237) and next causing neuroinfection. In fact, it was found that ACE2 and TMPRSS2 were expressed in the oligodendrocyte precursor cells and the astrocytes of the substantia nigra and cortex (82). COVID-19 can potentially damage the capillary endothelium within the brain and contribute to elevated blood pressure. The risk of SARS-CoV-2 cerebral hemorrhage through an ACE2 receptor can result in abnormally high blood pressure and increase cerebral hemorrhage. However, although ACE2 and TMPRSS2 are present in the nervous system, additional pathways were also hypothesized for the entry of SARS-CoV-2 into the nervous system, including the direct intranasal entry to the brain via olfactory nerves, the indirect entry to the brain go through the blood-brain barrier via hematogenous or lymphatic spread, the hypoxic injury, and finally the immune-related injury (7,258). It is known, that coronaviruses can enter to the nervous system straight through the olfactory nerve, potentially causing loss of smell and taste, and enter the nervous system through blood circulation and neuronal pathways. In addition, coronaviruses, including SARS-CoV-2, trigger harmful effects in the lung tissue leading to several lung lesions and consequent hypoxia, that can be responsible of the brain disease progression. These data highlighted that awareness, management and timely analysis of infection-related neurological complications of SARS-CoV-2 patients are key to improve the prognosis of severe ill patients. Skin On October 7, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND (skin OR cutaneous manifestation) we found 771 reports. Skin manifestations due to SARS-CoV-2 infection are of different types and currently reported in numerous case reports, case series, and literature reviews (259)(260)(261)(262)(263)(264). The first case study on skin manifestations was published by Recalcati et al. and included 88 patients that showed widespread urticaria, erythematous rush and chickenpox-like vesicles (265). Subsequently, other authors described urticarial rash petechial also in association with decrease platelet count and sometimes also with eosinophilia (265)(266)(267)(268)(269)(270). Zhang et al. evaluating 140 patients with SARS-CoV-2 infection, stated that urticaria were self-reported by 1.4% of patients (268). Despite, the majority of studies reported that urticarial skin manifestations were not correlated with SARS-CoV-2 severity (265,268), a prospective cohort study reported that the presence of urticaria and maculopapular skin lesions were associated with higher morbidity and higher mortality rate (2%) (271). In addition to urticarial skin manifestations, Manalo et al. also described a transient livedo reticularis as potential skin manifestation linked to SARS-CoV-2 (272). Other described skin manifestations are related to acral ischemia often related to an hypercoagulation status of SARS-CoV-2 patients, that have a negative prognostic implication in virus evolution (273)(274)(275). These manifestations could be caused by direct injury of vascular endothelium by SARS-CoV-2, which could lead to DIC, antiphospholipid syndrome, and vasculitis mimics. Case series showed purpuric skin involvement in severe SARS-CoV-2 patients, in detail retiform purpura on the buttocks, dusky purpuric patches on the palms and soles, and livedo reticularis on the chest and limbs were detected (261,273,276). Tissue biopsies from skin and lung detected thrombogenic vasculopathy and deposits of C5b-9 and C4d complement proteins (273). This was in line with widespread activation of both alternative and lectin pathways of complement, suggesting that severe SARS-CoV-2 patients can suffer thrombotic microvascular injuries that can involve not only the lungs but also the skin, and probably other organs (273). Skin manifestations were found also in pediatric patient where the skin lesions commonly happen in asymptomatic or mildly symptomatic children and adolescents (277)(278)(279). Skin biopsy of acral perniosis lesion in SARS-CoV-2 pediatric patients revealed a superficial and deep lymphocytic infiltrate, where vacuolar change and purpura were also present (280,281). Hemorrhagic parakeratosis in the stratum corneum were also detected and as well as dermal infiltrate strongly perivascular and perieccrine and lymphocytic vasculitis in the thin muscular walls of small vessels (4,205). Similar results were also found in skin biopsies from SARS-CoV-2 adult patients that showed a lymphocytic perivascular and perieccrine infiltrate (282,283). To date, there are still not enough studies to define which are the skin manifestations of SARS-CoV-2 infection, and why they occur. As reported by Recalcati et al. these dermatological manifestations "are similar to cutaneous involvement occurring during common viral infections" (265). Several hypotheses could be formulated from integration of the clinical observations and data from literature, but to date whether these skin manifestations were neurogenic, microthrombotic, or immune complex mediated is unclear. However, examine the tissue samples to understand if SARS-CoV-2 can be detected in the skin itself could be of key importance also considering that ACE2 is present in basal epidermal layers and in sebaceous gland cells of the skin (58). In addition, a recent study detected that ACE2 and TMPRSS2 were co-expressed at the epithelial sites of the skin, highlighting the potential roles of these molecules in SARS-CoV-2 (284). However, so far it is not known if skin manifestations (nonpruritic, erythematous rashes, urticaria, or varicella-like lesions) in COVID-19 patients are a place of viral replication or just a local reaction to systemic infection. Male and Female Reproductive System and Pregnancy On October 7, 2020 searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND (reproductive system OR ovaries OR testis OR pregnancy)" we found 1,301 reports. Most of these reports described high levels of ACE2 expression in the testes, spermatids, ovaries, fallopian tubes, placenta, and uterus, thus highlighting a potential high risk of SARS-CoV-2 infection in the human reproductive system (61,(285)(286)(287)(288). However, data on the presence of SARS-CoV-2 in male reproductive system are conflicting. A study carried out by Li et al. revealed that SARS-CoV-2 was found in the testes of infected cases (289). A post-mortem study on 91 COVID-19 victims also showed varying degrees of spermatogenic cell reduction and damage, and presence of SARS-COV-2 RNA and virus particles in the testes (290). Conversely, some clinical studies did not detect SARS-CoV-2 in semen or testicular biopsy of COVID-19 patients (291)(292)(293). Thus, it is possible to speculate that SARS-CoV-2 gains access to the male reproductive system in some but not in every COVID-19 patient. However, since SARS-CoV-2 can lead to systematic effect it can have also other consequences on the reproductive system. Several studies reported testicular discomfort and devastation to the testicular parenchyma in COVID-19 patients even when the testes were SARS-CoV-2 negative (291,293). As known, the reproductive health issues may not be restricted to men, but woman may also have consequences. What seems to be quite clear is the distribution and function of ACE2 in the female reproductive system. Jing et al. clearly reported the ACE2 expression in the ovary, uterus, vagina, and placenta (60). Moreover, since Ang II, ACE2, and Ang-(1-7) regulate follicle development and ovulation, modulate luteal angiogenesis, and degeneration, and influenced the regular changes in endometrial tissue and embryo development SARS-CoV-2 infection may disturb the female reproductive functions, resulting in infertility, menstrual disorder and fetal distress (60). Although these data suggested that there are potential routes for SARS-CoV-2 to compromise female fertility, currently no studies on damage to female COVID-19 patients' reproductive system were reported. For SARS-CoV-2 role in female reproductive system, the latest evidences were mainly focused on pregnant women. The simultaneous expression of ACE2 and TMPRSS2 seems to lack at the cellular level of maternal-fetal interface. Despite the clinical manifestation in COVID-19 pregnant women seems to remain the same as in non-pregnant patients, several studies suggest that pregnant women infected with COVID-19 may be at risk for preterm delivery (294)(295)(296). Recent papers also reported cases of pre-eclampsia and manifested gestational hypertension in COVID-19 pregnant women (297)(298)(299). An analysis of the WAPM study on COVID-19 reported that early gestational age at infection, maternal ventilatory supports and low birthweight are the main determinants of adverse perinatal outcomes in fetuses with maternal COVID-19 infection (300). However, significant neonatal respiratory diseases appear to be rare in presence of SARS-CoV-2 positivity (301). In this context, the key question is whether SARS-CoV-2 can be transmitted to fetuses from a woman infected with COVID-19. The evidence of infection in infants in the time immediately following birth (since few hours to a couple of days) suggests the possibility of maternal fetal transmission, via intrauterine vertical infection or mediated by breastfeeding. In this latter case, evidence of the presence of Sars-Cov2 in maternal milk is still controversial, and in the close contact between mother and child in these phases could lie the true way of transmission. Instead, despite primary reports from China suggested that vertical transmission was unlikely, several case series revealed the possibility of vertical transmission from positive SARS-CoV-2 woman (302,303). Two conditions are mandatory for transplacental transmission to be possible: (1) SARS-CoV-2 must reach the placenta and (2) ACE2 must be present in the placenta. Regarding the first condition, several papers supported the presence of SARS-CoV-2 in placental tissue. In particular, histopathological signs of placenta alteration have been observed in pregnant women affected by Sars-CoV-2, with evidence of inflammatory state and alteration in vascular supply. (304)(305)(306)(307)(308). Regarding the second condition controversial results are still present (309,310). However, a recent study indicated that trophoblastic cells, which are in direct contact with the maternal blood in the intervillous space, showed a strong expression of ACE2 throughout pregnancy, supporting that SARS-CoV2 is able to infect the placenta via a receptor-mediated mechanism (311). A further study investigated the potential transmission routes in the first trimester, and they found expression of ACE2 and co-expression of TMPRSS2 in the trophoblast, blastocyst, and hypoblast (312). However, other proteases such as Furin, trypsin and cathepsins B and L could be also implicated (16,313,314). Thus, despite the lack of clinical evidence, SARS-CoV-2 infection may carry a potential risk of reproductive system. Thyroid On October 7 2020 by searching on PubMed "COVID-19 OR COVID-2019 OR severe acute respiratory syndrome coronavirus 2 OR severe acute respiratory syndrome coronavirus 2 OR 2019-nCoV OR SARS-CoV-2 OR 2019nCoV OR (Wuhan AND coronavirus) AND thyroid" we found 112 papers. Data on direct thyroid involvement in SARS-CoV-2 infection arescarce and most of the reports are focused on identifying a possible association between hypothyroidism and outcomes related to COVID-19. A consensus statement regarding issues specific to thyroid dysfunction during SARS-CoV-2 pandemic was issued by the British Thyroid Association and the Society for Endocrinology (315). The consensus suggested to patients with hypothyroidism or hyperthyroidism to continue their medications, however, it underlined that patients on anti-thyroid drugs are at a risk of agranulocytosis, symptoms that often overlap with those of SARS-CoV-2 (315). However, recently, van Gerwen et al. evaluated 3,703 COVID-19 patients of which 251 patients (6.8%) had pre-existing hypothyroidism and received thyroid hormone therapy (316). They found that hypothyroidism was not associated with increased risk of hospitalization, mechanical ventilation, and death (316). A direct thyroid involvement associated with COVID-19 was highlighted by Campos-Barrera et al. that identified a subacute thyroiditis associated with a very mild presentation of COVID-19 in a healthy 37-year-old female (317). Subacute thyroiditis was not the only thyroid condition associated with COVID-19. In fact, cases of thyroxine thyrotoxicosis have been also described (318). Several case reports and a case series were focused on the prevalence of subacute thyroiditis and thyroxine thyrotoxicosis in patients with severe presentation of COVID-19 from ICU (319)(320)(321)(322)(323)(324). More recently in a retrospective study on 50 COVID-19 patients it was found a decrease in total T3 and TSH concentrations in 56% of patients (325)(326)(327). The decrease in T3 concentration resulted more pronounced in patients with the severe SARS-CoV-2 (325). Despite the few data related to the thyroid involvement during SARS-CoV-2 infection, it is important to emphasize that, as previously reported ACE2 expression levels were high in thyroid and its expression were positively and negatively associated with immune signatures in males and females (328). Additionally, TMPRSS2 was also expressed in thyroid (82). Therefore, surely further studies would be important to understand a potential involvement of the thyroid in SARS-CoV-2 infection. DISCUSSION Since it has been demonstrated that the novel SARS-CoV-2, which affected a very high number of people all over the world, entry into the cell exploiting ACE2, more and more research and studies are focusing their attention on ACE2 role, function, and distribution and on its interaction with specific proteases that assist SARS-CoV-2 infection. In fact, it is known that following the entry of the virus into the human cell through the binding with ACE2, the S protein is cleaved by TMPRSS2, with the help of Furin which facilitates the entry of the virus into the cell after binding. However, theoretically, also other human's proteases (cathepsin L and B, elastase, trypsin and factor X) could be involved in this complex process and numerous studies are currently ongoing. Our overview highlighted that ACE2 receptors, being ubiquitous, and extensively expressed in numerous human tissues and organs, such as in the heart, vessels, gut, lung, kidney, testis, and brain and many other, may play a key role in the involvement and subsequent impairments of various organs during the SARS-CoV-2 infection. ACE2 is typically bound to cell membranes and poorly present in the soluble form in circulation. In addition to its negative role in SARS-CoV-2 infection, and in other virus, membrane-bound and soluble ACE2 also perform beneficial biological functions, the main represented by the degradation of angiotensin II to angiotensin 1-7. Thus, ACE2 receptors cut down some harmful effects consequential to the bind of angiotensin II to AT1 receptors, which comprise vasoconstriction, increase inflammation, and thrombosis (329). However, the entry of SARS-CoV-2 in the cells by membrane fusion down-regulates ACE2 receptors, thus SARS-CoV-2 seems to entry into the cell with the membrane receptor, which is functionally detached from the membrane external site. This phenomenon can cause the detrimental effects in SARS-CoV-2 infection. It is important to underline that several other factors, such as genetics, demographic, lifestyle, co-morbidities and drugs usage could have a potential impact on ACE2 expression and activity. In fact, it was extensively reported that SARS-CoV-2 patients present several features associated with infection and severity of the disease, such as older age, hypertension, diabetes and cardiovascular disease, that share a different degree of ACE2 deficiency and that can produce bias in the evaluation of the effective damages caused by the virus (1). However, despite during SARS-CoV-2 infection ACE2-expressing organs may become direct targets, leading to critical pathological manifestations and subsequent multiple organ failure or even death, the exact mechanism and effective action through which ACE2 act on these organs is still heavily debated. Another point at the center of the clinical and scientific debate is represented by the potentially beneficial effect (or not) of soluble form of ACE2, the form that lacks the membrane anchor and circulates in small amounts in the blood. A paper by Battle et al. hypothesizes that the soluble form of ACE2 might behave like a competitive interceptor of SARS-CoV avoiding the binding of the virus to the surface-bound, full-length ACE2, the form that contains a structural transmembrane domain, which anchors its extracellular domain to the plasma membrane (330). This evidence is in line also with in vitro studies (331,332). A preclinical model of Vero-E6 cells, infected with SARS-CoV-2, isolated from a nasopharyngeal sample of COVID-19 patient, demonstrated the efficacy of human recombinant soluble ACE2 (hrsACE2) in inhibiting viral replication in a dose-dependent manner. Such activity was also confirmed in human capillary organoids cultures and in kidney organoids cultures generated from human embryonic stem cells (331). In addition, the soluble ACE2 form seems to be also involved in blocking SARS-CoV-2 replication and in immune response against the virus, in concert with Fc portion of immunoglobulin (331). The administration of rhACE2 also seems to induce a reduction of IL-6 levels in severe COVID-19 patients (333). The increased production of IL-6 and other inflammatory cytokines (IL-1β, IL-2, IL-7, IL-8, IL-10, granulocyte-colony stimulating factor, granulocyte macrophage-colony stimulating factor, interferon-inducible protein-10, monocyte chemotactic protein 1, macrophage inflammation protein-1α, IFN-γ, and TNF-α,2,3,12,15) together with the presence of lymphopenia, lymphocyte activation and dysfunction, abnormalities of granulocytes and monocytes, increased production of immunoglobulin G (IgG) and total antibodies in COVID-19 patients points out how the SARS-CoV-2 is able to disrupt also the normal immune responses, leading to an impaired immune system (334)(335)(336)(337)(338)(339). Lymphopenia is a key feature of patients with severe COVID-19 (334). A marked reduction in the number of CD4+ T, CD8+ T, NK, and B cell was detected in these patients (335). In addition, a high expression of CD69, CD38, and CD44 on CD4+ and CD8+ T cells was seen (336). Virus-specific T cells from severe COVID-19 patients also highlight a central memory phenotype with high levels of interferon (IFN)-γ, tumor necrosis factor (TNF)-α, and IL-2. Nevertheless, lymphocytes have an exhaustion phenotype with programmed cell death protein-1 (PD1), T cell immunoglobulin domain and mucin domain-3 (TIM3), and killer cell lectin-like receptor subfamily C member 1 (NKG2A) upregulation (337). Unlike eosinophils, basophils, and monocytes percentage that were reduced in severe COVID-19 patients the level of neutrophils resulted increased (338). Thus, the damage and inefficiency of the immune system caused by lymphopenia, T cell exhaustion and cytokine release syndrome, and organ specific ACE2 expressing cells (endothelial, alveolus in lungs, proximal tubule, and glomerulus in kidneys, pericytes in heart, ect) could potentially lead to complications like acute respiratory disease syndrome and multi-organ failure. These complications not only can lead to a poorer outcome to the SARS-CoV-2 infection but can also lead to permanent alterations that can persist long after viremia (long-term COVID-19), such as pulmonary fibrosis, neurodegenerative deseases, cardiovascular and kidney diseases (340,341). Highlighting the pathological basis and mechanisms of COVID-19 and all the functions and activities of ACE2 during the virus would be essential for our understanding of the pathophysiology of the disease. A great help to our understanding could come from pathological studies of larger series of autopsy findings. Furthermore, the development of advanced and alternative preclinical models could help to discover more about the SARS-CoV-2 infection process itself, to analyze specific aspects of ACE2 in relation to SARS-CoV-2 pathophysiology and, most importantly, to learn the disease progression pattern observed in humans. In addition, more exhaustive and systematic studies on the physiological localization and activity of ACE2 might help in the comprehension of the mechanisms underlying the infection. In this regard, attention should be paid to the investigation of the cells of the immune system, in consideration of preliminary scientific evidence on the identification of ACE2 in immune cells residing in the tissues. This could open further scenarios on both the virus spreading mechanisms and tissues damage. We believe that devote scientific efforts for the clinical management of SARS-CoV-2 patients, also considering a personalized strategy aimed to provide individually tailored treatment for each patient, are currently mandatory. As showed in this report this aspect should also considered specific patient differences in the mutual interactions ACE2-SARS-CoV-2 with their consequences for the disease pathophysiology. Another interesting aspect that could be explored in patients who have overcome the disease is the possible onset or persistence of the alterations above described in the organs and systems and the evaluation of whether they are transient or permanent (long-term COVID- 19), to assess the extent of ACE2 activity impairment due to SARS-CoV-2 infection. AUTHOR CONTRIBUTIONS FS, MF, and ML designed the manuscript. FS and MM collected and analyzed literature, wrote the manuscript, edited, and prepared manuscript for submission. ML and MF revised the manuscript. All authors read and approved the final manuscript. FUNDING This work was supported by grants from IRCCS Istituto Ortopedico Rizzoli (Ricerca Corrente).	2020-12-03T14:11:27.917Z	2020-12-03T00:00:00.000	{ "year": 2020, "sha1": "e7ca693ba549185553a76866ef9590369750f482", "oa_license": "CCBY", "oa_url": "https://www.frontiersin.org/articles/10.3389/fmed.2020.594495/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "e7ca693ba549185553a76866ef9590369750f482", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
119454871	pes2o/s2orc	v3-fos-license	Distances to Nearby Galaxies: Combining Fragmentary Data Using Four Different Methods The primary distance indicators are established in our Galaxy and the Local Group. There are at least four different methods which give good distances: methods using proper motions, RR Lyraes, Cepheid variables, and Type II supernovae. However the data on independent distances is very fragmentary, due partly to nature and partly to technological limits. As a result the data are rarely put together in a consistent way; instead, the discussion of distance scales is often focused on one or two methods or on individual objects. Hence the question: what is the current situation with our overall knowledge of distances to the nearby galaxies? We try to answer this question by combining the fragmentary data from all four methods for fifteen objects: the galactic center, the globular clusters M2, M3, M4, M13, M22, M92, and 47 Tuc, the galaxies IC1613, M31, M33, M81, M100, and M101, and the Large Magellanic Cloud. We pay special attention to covariances among the different distance estimates. This most complete combination to date shows that all four methods are consistent within their random and systematic uncertainties. Introduction The demand for accuracy in primary distance indicators, like the Cepheids and RR Lyrae stars, has increased steadily in recent years. Nevertheless, certain problems have remained unresolved and are a cause for concern { the systematic o set between Cepheid and RR Lyrae distances, and the issue of Cepheid zero point dependence on metallicity, among them (Walker 1992;Gould 1994;van den Bergh 1995). The present paper is an e ort to bring together for comparison and scrutiny four basic methods, drawing widely upon the literature. In particular, we try to lift some of the degeneracy in the Cepheid vs: RR Lyrae problem by introducing completely independent methods { proper motions and supernovae. In this respect, we take advantage of the large amount of proper motions data now available as a result of the concerted e ort of K. M. Cudworth, R. J. Rees, R. C. Peterson and associates on globular clusters (Cudworth 1986;Rees 1992;Peterson et al. 1995). The data on independent distances has a highly fragmentary character. Some of the incompleteness is natural { old systems, like globular clusters, have no Cepheids or Type II supernovae; most of the incompleteness is technical { the nearby galaxies are too far for our ability to measure proper motions. We see our task as that of constructing the Ei el Tower shown on the frontispiece of the Decalages vers le rouge et expansion de l'univers proceedings (Balkowski & Westerlund 1977); we use four methods to support our structure, but there are missing parts, joints, even levels. On top of it, of course, is H 0 . Therefore we look for consistency in the comparison from the whole sample, rather than paying more attention to individual well-studied objects. Our goal is to study in a comprehensive but consistent manner the distribution of the distances these four methods provide to 15 objects in the vicinity of our Galaxy, including the galactic center GC]. We have no strong preference for any particular calibration or correction. In the rst section we describe brie y each of the four methods for distance determination. The next section has the details of their application to each of the 15 objects. This is followed by a discussion of the methodology and analysis of the results, ending with a summary. Combining Data from Di erent Methods Before we describe each of the methods used, we outline our approach. We have n galaxies and clusters for which we obtain distances. We shall represent those distances by an n-dimensional distance vector with an n-dimensional uncertainty ellipsoid. The coordinates of the center of the uncertainty ellipsoid will be best estimates of the distances to the objects. The uncertainty ellipsoid de nes the error range of the distance vector. { 3 { Distance measurements are often correlated; we shall use covariance as an approximation of correlation between di erent measurements. The uncertainties of the distance measurements will be then represented by an uncertainty matrix , where the o -diagonal elements, ij 2 , are the covariance terms. The inverse of this matrix, the weight matrix W = 1 , will give the coe cients of the equation of the uncertainty ellipse, or in our n-dimensional case { of the n-dimensional ellipsoid. The average distance vector, v ave , will then be: v ave = ( Wv; (1) where v is the n-dimensional distance vector and N is the number of methods for estimating distances (we consider four, N = 4). Dealing with Incomplete Data The application of equation (1) is straightforward when there are measurements for all systems using all methods. But when a method has not been applied to one of the systems the corresponding element of the vector v is unknown. We take the expedient approach of choosing a value not unlike the other values for that system, but with a very large uncertainty. We adopt this as a bookeeping device, a placeholder, or what we prefer to call { a \wild guess". Thus, it is a \wild guess", yet perfectly correct, to estimate that your computer keyboard is (50 45)cm wide. Similarly, if Cepheids give us a distance modulus to M31 of 24:5mag, we can safely say that the proper motions distance modulus to M31 is (20 10)mag, although it has never been measured. In the statistical analysis the \wild guesses" will have vanishingly small weights with as large fractional error estimate as 10mag { they contribute very little to 2 . Also, we took the correlation between the \wild guesses" and measurements to any other galaxies to be zero as well for the sake of consistency and symmetry. Choosing Data In choosing data we used the following criteria: (1) absolute, not relative distances; (2) availability of more or less homogeneous distances; and (3) published results, not samizdat. We were willing to use preliminary results from conference proceedings, but we refrained from using (better quality) data which had been privately communicated but not yet published. { 4 { Our use of very incomplete methods perhaps needs explanation. Our reasoning is that any measurement using a new method adds additional information (and an additional check) which we are loath to forego. Proper Motions and Kinematic Distances Proper motions provide the simplest method, although they work only for the objects which are su ciently close { ten out of fteen objects that we examined have had internal proper motions measured. The basic idea is to measure the radial velocity v of the star, which, with suitable assumptions about isotropy, is a good approximation of the star's tangential velocity. The relationship between the tangential and angular velocity is d = v w where d is the distance that we want to obtain. Since the measurements of proper motions in di erent systems are uncorrelated, we have set all the covariances, ij , equal to zero. This assumes that there is no systematic error associated with the assumed isotropy of radial and transverse velocities. RR Lyraes For the method of RR Lyrae variable stars, we have taken apparent magnitude m (with its uncertainty m) as measured and correct it for reddening using the familiar term RE(B V ) and adopting for the ratio of total to selective absorption R = 3.1, as suggested by many authors. (In some cases the correction for reddening was already made by the original author(s) and we didn't have to make it). We calculated the absolute magnitude for RR Lyraes using the Carney et al. (1992a) relation M = 0:15( 0:01) Fe=H] + 1:01( 0:08)(mag); (2) which gives the absolute magnitude M as a linear function of metallicity Fe=H]. With a little calculation it is easy to see that the the biggest error in this equation is that of zero point, and that two other errors, ones from the slope term and metallicity term are much smaller and can be ignored. We therefore adopted M = 0:08 mag. Now we can calculate the mean distance modulus as DM = m 0 M. Since we have m (random statistical error in m) and M (systematic error in M), we can calculate the total uncertainty in distance modulus according to the usual formula for propagation of uncertainties DM = q ( m) 2 + ( M) 2 : Calculations of distances using RR Lyraes have their errors scattered randomly, except for Carney's equation which is used with every object and which \pulls" the error in the same direction every time. Therefore, we have chosen to take ij = Carney = 0:08 as the covariance between any two objects. We should note that Walker (1992) suggests a calibration, based largely on SN 1987A, and gives M = 0:15 Fe=H] + 0:73 (mag) for which the zeropoint is almost 0:3 mag brighter than Carney's zeropoint. However, more recent results seem to be closer to Carney's equation (Walker 1995). Cepheids For Cepheids, we have used the set of Caldwell & Coulson's (1987; henceforth CC) PL-V and PL-I equations (in the BVI C system). We did not use their PL-C relation, because it needs B-photometry in a signi cant way; such photometry is not available for any of the HST Cepheid distances which we want to use. The equations for the PL-V and PL-I cases are M PL V = 3:11 logP 3:77 and m PL V = hV i 0 ; hV i 0 = hV i R V E(B V ); (7) and hI C i 0 = hI C i R I C E(B V ): (8) and E(B V ) is reddening, P is Cepheid's period of oscillation (in days); magnitudes with subscript \0" have been corrected for reddening. Here R B , R V and R I C are ratios of total to selective absorption for the three bands respectively. We have adopted R V = 3:1 as used for most of our RR Lyrae distances. Following Taylor (1986) we obtain R I C = 1:77, which includes the small term for G supergiants suggested by the author. We use the corrections for metallicity e ects from CC in terms of depletion, de ned as d = 1 10 Fe=H] : For example, in the PL-V case this is the metallicity dependent bolometric correction B:C: = (0:134BV 0 0:046)d { 6 { where BV 0 = hBi hV i E(B V ): (11) Whenever B-photometry was not available, we used a relation with V-I from Caldwell & Coulson (1985). The question of metallicity dependence trends in the Cepheid distances requires special attention (see Stothers 1988, andGould 1994). We tried to test for metallicity dependence in the Cepheid distances in two di erent ways. First, by comparing them to the distances measured with RR Lyrae stars and minimizing the di erences with a linear dependence on Fe/H]. Second, by using the average distance ellipsoid of the entire sample to form these di erences (see x4). In both cases we nd a similar weak dependence on metallicity with a questionable goodness-of-t to it. We took data for log P, hBi, hV i, and hI C i for a number of Cepheids in each object and calculated absolute and apparent magnitude for the PL-V, and PL-I case for each Cepheid separately. Whenever the observational data was in the Johnson system, we used the empirical relation obtained by CC from 45 Cepheids: I C = I J + 0:272(V I) J 0:049; which is in good agreement with the general transformation in Taylor (1986). This concerns only the PL-I case. Then we obtained distance moduli DM PL V and DM PL I for each Cepheid. It was then easy to calculate average distance moduli for PL-V and PL-I case as the arithmetic means of all the corresponding distance moduli and their corresponding random uncertainties as where DM is the mean distance modulus and K is the number of Cepheids in a galaxy that we considered. Now we have PL-V and PL-I distance moduli with their corresponding random uncertainties. For a couple of our objects these moduli are almost identical, while the discrepancy seen in the rest is insigni cant (x3). Therefore, we took the arithmetic mean as the nal distance modulus and the arithmetic mean of the two uncertainties as the nal random uncertainty, accounting for the above systematic di erence in the uncertainties. We did not use the formula for propagations of uncertainties because PL-V and PL-I distance moduli are not independent at all. In our use of the CC PL-V and PL-I relations we have xed the extinction laws (through the values of R V and R I C ). However we have relied on Cepheid reddenings derived by Freedman et al. (1994a) using a method which is di erential { 7 { with respect to the LMC and may apply a di erent R I C . We are not able to identify any origin for a systematic di erence between the PL-V and PL-I distance moduli (see also Madore & Freedman (1991)), and only note that if any, the systematic is insigni cant given the random uncertainties. Feast (1991) notes that the CC distance scale is tied to van Leeuwen's parallax for the Pleiades, which is uncertain by 0:08 mag. Indeed, a preliminary Hipparcos parallax for the Pleiades reported by M. Penston in the February 1994 Observatory is 0:09 mag more distant. Therefore, a good estimate for the covariances in the Cepheid method is ij = 0:08mag for any two objects. The nal uncertainty is combination of random and systematic (covariance) uncertainty and we can calculate it using the usual formula for propagation of uncertainties DM final = q DM 2 mean + 2 ij : In some systems Caldwell & Coulson gave a nal uncertainty. However, while they used ij = 0:03 as their systematic error, we prefer ij = 0:08, leading to the following correction DM final = q DM 2 CC 0:03 2 + 0:08 2 : Type II Supernovae The expanding photosphere method for SNe II was introduced by Kirshner & Kwan (1974) and is a variant of the Baade-Wesselink method used in radially pulsating stars (Baade 1926;Wesselink 1946). The use of a distance correction factor derived from models of the expanding atmosphere (via ts to the synthetic spectral distributions) seems to divide the derived distances into shorter (Eastman & Kirshner 1989, Schmidt et al. 1994) and longer scales (Branch 1987, Baron, Hauschildt, & Branch 1994. The former scale is based on a physical, albeit simplistic, assumption of a two-level approximation, while the latter scale is an ad hoc corrected upwards parameter t. We prefer to use the former scale. Remaining Inhomogeneities Homogeneity and consistency are our main objectives in this study. We used each of the methods for deriving distances according to the prescriptions described above. Thus we avoided the use of distances derived by the various authors in order to retain control over the assumptions and uncertainties which enter into each distance determination. In a { 8 { couple of cases we used data of inferior quality for the sake of preserving overall consistency. A case in point is our use of the optical V-band RR Lyrae relation for M4, instead of the superior new K-band relation which minimizes extinction problems. Despite our e orts, the fragmentary character of the data on independent distances leaves us with some remaining inhomogeneities. In the kinematic distances the main problem with inhomogeneity arises because some of the proper motions and radial velocity data is not readily available. For example, we use a distance to M4 from Peterson et al. (1995) which is based on dynamical models by , while our distance to GC is simply d = v w , where v and w come from separate studies. The kinematic distance to M13 is also model dependent (Lupton et al. 1987). Finally, the kinematic distance to the LMC is based on a very di erent approach altogether. All this is of no serious concern to us because all these kinematic distances share a common feature: they are completely independent from the other methods. Therefore our overall comparison of the four methods remains meaningful and consistent. In the RR Lyrae distances there is one main inhomogeneity. In the globular clusters M2, M3, and M13 we used an extrapolation of the blue horizontal branch sequence, because of the lack of RR Lyrae stars or of their photometry. This reduces the precision but does not introduce a new systematic error. In the Cepheid distances there is no particular inhomogeneity. In the SNe II distances the inhomogeneity is caused by the fact that the SN 1987A in LMC and SN 1993J in M81 were not typical Type II supernovae and their distance correction factors may be systematically di erent (Schmidt et al. 1994). Distance Calculations 3.1. The Galactic Center Mould (1983) measured radial velocity for 49 M giant stars at the galactic center and obtained a dispersion v = (113 11)km=s. Spaenhauer et al. (1992) Blitz & Spergel (1991)), the bulge of the Milky Way is bar-like, our implicit assumption of isotropic orbits may not be correct. Walker & Mack (1986) give, among other interesting results, complete data for 6 RR Lyrae stars in Baade's window near the Galactic Center (Table 1). Since we have hV i(= m V ), Fe=H] and E(B V ) for each star, we can calculate their distance moduli separately, as described in the previous section. The mean value of those 6 distance moduli is (14:29 0:08) mag. Taking into account the uncertainty (systematic error) in Carney's relation, we have total = q 2 mean + 2 Carney = 0:11 mag, giving DM = (14:29 0:11) mag. As regards a Cepheid distance, Caldwell & Coulson (1987) give us PL-C, PL-V and PL-I distance moduli for the GC which they obtained using the procedure explained in the previous section including all corrections. As noted above, we use only their PL-V and PL-I cases. Since they give DM PL V = (14:42 0:18)mag and DM PL I = (14:40 0:18)mag, giving a mean distance modulus DM = (14:41 0:18) mag. Correcting it for the covariance of 0:08 mag (as explained in the previous section), we nd DM = (14:41 0:19) mag: An implicit assumption in CC's distance is that orbits are roughly circular. If the potential of the Milky Way is as elliptical as Kuijken & Tremaine (1994) claim on the basis of the local velocity ellipsoid, CC have overestimated the distance. Adding a new distance indicator We decided to include the new and widely recognized method of measuring proper motions in the Galaxy (the water maser (WM) method) and see its impact on our average { 10 { ellipsoid. Unfortunately, only one distance is available { the distance to the GC. Reid (1993) gives d = (7:2 1:3)kpc which is equivalent to DM GC = (14:29 0:39)mag. All the other distances in the WM distance vector will be our \wild guesses". Since Reid also speci es that the systematic error in the measurement to GC is 1:1kpc, we adopt, after conversion to magnitudes, that xy = 0:33mag for any two elements in the WM uncertainty matrix. After looking at any graph which contains the WM method or calculating the coordinates of a center of our new average distance ellipsoid, we can conclude that the impact of WM is absolutely negligible. The explanation is that the WM measurement has a very large uncertainty, and therefore, it has very little contribution in forming the average distance ellipsoid. M 4 This is a low concentration globular cluster which is relatively near, but is heavily reddened by the Sco-Oph dark nebulosity with unusual extinction properties. Most recently the RR Lyrae variables in M 4 have been studied by Liu & Janes (1990) { we used the light curves and individually determined E(B-V) for four of them. The metallicity of M 4 is Fe=H] = 1.05 (Drake, Smith, & Suntze 1994), which con rmed independently the derivation of fundamental parameters for M 4 by Dixon & Longmore (1993), whose R = 4:0 0:2 we adopt. The mean value of the four distance moduli and the systematic error in Carney's equation accounted for gives us nally DM = (11.07 0.11) mag. M22 The globular cluster M 22 is similar to M 4 but farther away. It is similarly heavily reddened, but more uniformly and with a seemingly normal extinction law. Peterson & Cudworth (1993) have measured proper motions of 672 stars in the eld of M22, and thus obtained the mean distance d = (2:57 0:29)kpc, or DM = (12:05 0:23)mag. Webbink (1985) gives m = 14:20 mag, based mostly on photographic light curves of RR Lyraes by Wehlau & Hogg (1978), who preferred the value m = 14:1 mag. We adopt the latter, as it agrees very well with the CCD photometry of M 22 by Anthony-Twarog et al. (1995). A reasonable guess for the uncertainty here is m = 0:09mag, which accounts also for the large uncertainty in the reddening { E(B V ) = (0:42 0:03) by Crocker (1988). Lehnert et al.(1991) estimate Fe=H] = 1:54. Now we have all the necessary data and, using the same procedure as before, we nd DM = (12:02 0:12)mag. There is only one RR Lyrae star, V9, in this globular cluster. There are no Cepheids and supernovae in 47 Tuc, so we just say that DM = (12:00 10:00)mag. M92 This is a very well studied metal-poor globular cluster with almost no reddening along the line of sight. Rees (1992) gives the distance obtained by measurement of proper motions d = (8:3 1:6)kpc which is equivalent to DM = (14:60 0:38)mag. Large Magellanic Cloud There are no internal stellar proper motions measurements in the LMC, but the distance to SN 1987A measured using its ring is derived in the spirit of astrometric measurements. We include it below and thus the LMC is the only object in our study which has its distance derived with all four methods. Panagia et al. (1991) calculated the distance to SN 1987A by comparing the angular size of its circumstellar ring with its absolute size obtained by another method. From there, they derived the distance to LMC to be DM = (18:50 0:13)mag. Recently Gould (1995) reexamined the measurement of the caustics in the ionized-emission light curves of the ring and derived an upper limit to the distance to LMC of DM = (18:37 0:04)mag, assuming also that SN 1987A lies 500 pc in front of the LMC center. If adopted as a distance, rather than an upper limit, it halves the uncertainty in the average (from all four methods) LMC distance. However, the average distance changes only slightly (by 0.007 mag), and none { 13 { of the other distances in our sample are a ected; the total 2 value in the comparison of all four methods shows only a slight improvement. Therefore, in order to retain our uniform approach, the Panagia et al. distance is used throughout. This is not to diminish the importance of the unprecedented precision of Gould's distance, but to illustrate the \egalitarian" nature of our approach. Walker (1992) gives data for 7 clusters with about 180 RR Lyrae stars total. We proceed in completely the same way as in the case of GC: we calculate distance moduli for the 7 clusters and their mean with its own statistical uncertainty (Table 2). Then we incorporate the uncertainty from Carney's equation and nally obtain that DM = (18:23 0:09)mag. For the Cepheids, Caldwell & Coulson (1987) nd, using the same calibration as for GC, that the best estimate of distance to LMC is 18:45mag. They do not give the uncertainty here, but they give the uncertainty in their paper from 1986 (where they say that DM = (18:65 0:07) is the best estimate, using a di erent calibration). After making correction for the systematic error of 0:08mag instead of 0:03mag we obtain DM = (18:45 0:10)mag: The distance derived to SN 1987A in the LMC with the Type II Supernovae method is 49 6kpc (Eastman & Kirshner 1989) or DM = (18:45 0:28)mag: IC 1613 IC 1613 is too far away for its stars to have its proper motions measured, so we adopt a \wild guess" DM = (23:00 10:00)mag . Saha et al. (1992) give us the RR Lyraes mean apparent magnitude using Gunn's green lter, hgi = (24:90 0:10)mag. In the same paper we nd the foreground absorption { 14 { for the green band which is A g = 0:07, and we get the corrected apparent magnitude hgi 0 = 24:83 mag. Since Carney's equation gives M V , we need to convert hgi 0 to the visual apparent magnitude hV i 0 . We do it with the help of the transformation by Kent (1985): g = V + 0:41(B V ) 0:19, which requires knowledge of (B-V) when only one of Gunn's lters has been used. Fortunately, a good estimate of the intrinsic color of the RR Lyrae stars is possible on the basis of their periods and metallicity. We obtain a range (B-V)=0.33 0.39, corresponding to a small uncertainty in the transformed hV i of 0.006 mag. For the metallicity of RR Lyrae stars in IC 1613 we use Fe=H] = 1:6, following the suggestion by Saha et al.(1992) and the metallicity estimate for the red giant branch by Freedman (1988). Finally, we have hV i 0 = (24:88 0:10)mag, and using our procedure for RR Lyraes, we obtain that DM = (24:11 0:13)mag. We take the complete BV RI data for 10 Cepheids in IC 1613 from Freedman (1988). We adopt Fe=H] = 1:3 (which gives a depletion d = 0:95), and E(B V ) = 0:04, both from Freedman's paper. Then we use the CC calibration as described above and get individual PL-V and PL-I distance moduli. The mean PL-V and PL-I distance moduli are DM PL V = (24:35 0:16)mag and DM PL I = (24:35 0:18)mag and our average distance modulus is DM = (24:35 0:17)mag. When we include the systematic error of 0:08mag, we get that the nal distance modulus is DM = (24:35 0:18)mag: No supernovae have been observed in IC 1613, so we adopt a \wild guess" DM = (23:00 10:00) mag. 3.11. M33 Greenhill et al. (1993) were successful in measuring proper motions for ve maser components in M 33 based on VLBI maps. The dispersions in transverse and radial velocities for the maser features provide a measure of consistency rather than a formal estimate of distance, which is d = (600 300)kpc, or DM = (23:9 1:5)mag. We used complete data for 11 Cepheids in M33 from Freedman et al. (1991). In the same paper we can nd that E(B V ) = 0:10, and a reasonable estimate for the metallicity Fe=H] = 0:3, in view of the abundances for 42 supernova remnants in M 33 (Smith et al. 1993). Now, using our procedure for Cepheids, we get PL-V and PL-I distance moduli for each Cepheid. The mean PL-V and PL-I distance moduli are DM PL V = (24:90 0:11)mag and DM PL I = (24:87 0:10)mag, and the average distance modulus is, therefore, DM = (24:88 0:11)mag. The nal distance modulus is, after taking into account the systematic error, DM = (24:88 0:14)mag: No supernovae have been observed in M 33, so we adopt a \wild guess" DM = (23:00 10:00) mag. M31 Since M31 has no kinematic distance, we adopt DM = (23:00 10:00)mag. For a RR Lyrae distance Pritchet & van den Bergh (1987) We took Cepheid data for log P, hBi, hV i and hIi from the graphs given for three of Baade's Fields ( elds I, III and IV) in Freedman & Madore's (1990) paper. In the same paper, we nd that the metallicities ( Z Z ) are 1.70, 1.14 and 0.30 and the reddenings E(B V ) are 0.20, 0.25 and 0.00 for the three elds respectively. Then we calculated average PL-V and PL-I distance moduli for each eld using the CC equations (Table 3). We took the average (arithmetic mean) of the 6 distance moduli that we thus obtained and the average of their uncertainties (we did not make any mistake by doing the latter thing, because the uncertainties were almost the same). The average distance modulus thus obtained is DM = (24:38 0:13) mag. After taking into account the systematic error of 0:08 mag, we get the nal distance modulus DM = (24:38 0:15) mag. No supernovae have been observed in M 31, so we adopt a \wild guess" DM = (23:00 10:00) mag. M81 With no kinematic distance measured to M81 and no RR Lyrae stars known there, we adopt DM = (27:00 10:00)mag as our \wild guesses". We used data for the 25 Cepheids found and observed (Cousins V and I) by Freedman et al. (1994a) with the HST. Metallicities for the M81 elds are only available from studies of HII regions (Garnett & Shields 1987;Zaritsky et al. 1994) and give Fe/H] 0.05. The reddening is estimated by Freedman et al. (1994a) to be E(B-V)=0.03. Then we calculated average PL-V and PL-I distance moduli from all the Cepheids using the CC equations, as applicable to VI data. The mean distance moduli are DM PL V = (27:98 0:09)mag and DM PL I = (27:95 0:08)mag and our average distance modulus is DM = (27:97 0:12) mag. When we include the systematic error of 0:08mag, we get that the nal distance modulus is DM = (27:97 0:14) mag. M101 Since M101 is too far away for proper motions observations and no RR Lyrae stars are known there, we adopt DM = (29:00 10:00)mag as our \wild guesses". We used data for the 4 Cepheids found and observed (BVRI photometry) by Alves & Cook (1995) at KPNO. Metallicities for the M101 elds are only available from studies of HII regions (Zaritsky, Elston, & Hill 1990), and are transformed by Alves & Cook assuming a solar Fe/O] ratio to give Fe/H] = 0.28. The reddening is estimated by the authors to be zero. Then we calculated average PL-V and PL-I distance moduli from all the Cepheids using the CC equations. The mean distance moduli are DM PL V = (29:26 0:17)mag and DM PL I = (29:45 0:19)mag and our average distance modulus is DM = (29:36 0:25) mag. When we include the systematic error of 0:08mag, we get that the nal distance modulus is DM = (29:36 0:27) mag. M100 { 17 { Since M100 is too far away for proper motions observations and no RR Lyrae stars are known there, we adopt DM = (30:00 10:00)mag as our \wild guesses". We used data from the HST observations of Freedman et al. (1994b) for 20 of their Cepheids in M100. These time-averaged intensity-weighted V and I magnitudes, kindly provided to us by Freedman (1995, private communication), are preliminary. As a metallicity estimate for the M100 Cepheids we adopt a value of ( Z Z ) = 1.25, derived from the abundances of HII regions (Zaritsky et al. 1994), assuming a solar Fe/O] ratio. The reddening is estimated by Freedman et al. (1994b) to be E(B-V)=0.05. We calculated average PL-V and PL-I distance moduli from all the Cepheids using the CC equations, as applicable to VI data. The mean distance moduli are DM PL V = (31:23 0:14)mag and DM PL I = (31:31 0:14)mag and our average distance modulus is DM = (31:27 0:20) mag. When we include the systematic error of 0:08mag, we get that the nal distance modulus is DM = (31:27 0:22) mag. Graphical Representation Each of the 4 methods we used gives a set of fteen distances and uncertainties which can be represented as an ellipsoid in a 15-dimensional space. Then we made projections of each such ellipsoid (including the average distance ellipsoid) onto the di erent planes de ned by the axes with distances to the objects. We show the projections for ve pairs of objects in Figures 1 and 2. We derive these projections in the following manner. For a given pair (x,y) of objects (and a given method) we create a 2-dimensional distance vector and a 2 2 uncertainty matrix by taking the i-th and j-th element from the 15-dimensional vector v, and by taking the elements ii , jj , and ij from the 15 15 uncertainty matrix (see x2. and Eqn.1). We thus form the uncertainty matrix xy = " 2 x 2 xy 2 xy 2 y # where x is the uncertainty in the x direction, y is the uncertainty in the y direction, and 2 xy is a covariance term. Then, from the general equation of the uncertainty ellipsoid v T (Wv) = 1; { 18 { the 2-dimensional weight matrix W = 1 xy = " p q q r # will give us the coe cients of the equation of the uncertainty ellipse px 2 + 2qxy + ry 2 = 1 and it is plotted on Figures 1 and 2. When dealing with a higher-dimensional con dence region a projection, not intersection, is used to a lower-dimensional space (e.g. Press et al. 1992), and to the extent that the con dence region is approximated by an ellipse on the plane, our projections are a relevant illustration to the analysis in the next section. Observing ve of them in Figures 1 and 2, we can o er the following general remarks. First, we note that all the average ellipses (projections of the average ellipsoid) are smaller than the original ellipses and are centered between them, which we expected to be the case. Second, we note that the \wild guesses" are represented by ellipses which are so elongated in corresponding (uncertain) directions that they look almost like two parallel lines (Fig. 1b,c). If the distance in both directions is a \wild guess", then we do not see the ellipse at all (Fig. 1a), because it is extremely big (with axes 10 mag long in each direction). All the distances (components of distance vectors) contribute in forming the components of the average distance vector, but their contributions are not the same. The limiting factor in distances' contributions are their corresponding uncertainties; distances with big uncertainties are almost \ignored" in forming the average distance, while those with small uncertainties play a major role there. Furthermore, if we make a change in the distance to one of the galaxies, all the distances of the average vector will be a ected, not only the component in the direction of that galaxy (although that component will be a ected the most, of course). The reason for that is that all the distances are correlated (except the proper motions distances), so the change in one direction will imply the change in all the other directions. 2 and some additional comments We can compare the obtained distributions of distances in several ways. First, we combine all 15 objects with their distances and uncertainties obtained with 4 methods for measuring distances, \wild guesses" included, and derive a best estimate, v ave , as already de ned by Eqn.(1). We compare the distributions of all three methods to their average by calculating Fig. 1.\| Projections of the distance ellipsoids for four objects of our sample. The ellipses due to each method are marked, e.g. KD is for kinematic distance; the small unmarked ellipses are the projections of the 15-dimensional average distance ellipsoid. There are no classical Cepheids in globular clusters, therefore \wild guesses" were used (see text), hence Cepheid distance ellipses are very elongated or not seen at all in such cases. The panels are marked (a) to ( Fig. 2 provides a limited illustration of this, as well. An alternative method which avoids the use of \wild guesses", is to combine objects in two groups: kinematic distances and RR Lyrae, and RR Lyrae and Cepheids (see Table 4). In each group we derive a corresponding new v ave and follow the procedure above (here the degrees of freedom will equal the number of objects in each group). In the comparison of RR Lyrae and Cepheids we have 5 objects and nd 2 = 4:5 for 5 degrees of freedom. For the combination of kinematic distances and RR Lyrae we have 9 objects and nd 2 = 4:0 for 9 degrees of freedom. A similar comparison between Cepheids and SNe II distances involves only 4 objects and gives 2 = 6:6 for 4 degrees of freedom. Finally, we can compare Cepheids and kinematic distances, alas for 2 objects only, and nd 2 = 0:31 for 2 degrees of freedom { Fig. 2 is perhaps more informative in this case. A di erent way to present our approach here is to compare directly the distance distributions of the pairs, accounting for their combined uncertainties. The outcome is necessarily the same; we do it for illustration. Then for the comparison of RR Lyrae and Cepheids we will have: 2 RR;CE = (v ce v rr ) T (W ce;rr )(v ce v rr )]; (17) where W ce;rr = ( ce + rr ) 1 . The number of degrees of freedom and 2 values are the same as above. Finally, we wanted to test the e ect of inclusion of \wild guesses" in our analysis of the whole set: we repeated the above pair comparisons using all 15 objects in all cases. The results were consistent with the above ones to within a few percent in the 2 values. This con rms the usefulness of adopting \wild guesses" in a problem with scarce and fragmentary data, in order to take advantage of the entire sample and account fully for statistical dependencies. We nd a hint for a discrepancy in the RR Lyrae distances, with respect to correlated (Cepheids), as well as uncorrelated distance measurements. This is seen also on several of the graphical representations (Figs.1-2): RR Lyrae ellipses give smaller distance moduli ( 0:1 mag) than the average ellipse. A possible reason for this slight disagreement of the RR Lyrae method with the other two could be the zeropoint in Carney et al. (1992a) equation for the absolute magnitude as a function of metallicity (see x2.5). Conclusions To summarize, we analyze the most complete combination of Local Group distances to date. We compare four basic distance indicators by applying them to 15 objects { 7 globular clusters, 6 nearby galaxies, the galactic center, and the LMC. The globular clusters have no Cepheids and supernovae; ve galaxies have no proper motions measured. All distances derived by the four methods to each of the 15 objects are compared and an average distance { 23 { vector in this 15-dimensional space is derived. All values are listed in Table 4. We tested in several ways the statistical consistency of the distance distributions among each other. Our conclusion from the 2 -tests and from the evidence in the projections of the distance ellipsoids is that all four methods provide distances which agree within their random and systematic uncertainties. Finally, we o er a few comments. The point of our paper is in the consistent combination; the above conclusion comes from it. Pairwise comparisons of the methods show discrepancies, some of which had been well known. First, there remains a systematic discrepancy between the Cepheids and RR Lyrae, in that the latter distances are smaller. This needs to be combined with what appears to be a very good agreement between the kinematic and Cepheid distance measurements. While the former discrepancy is well known, the latter agreement is a new result, in the sense that there is only one object (the GC) where we could compare the Cepheid and kinematic distance scales truly directly; instead, here we could infer agreement from the whole sample. However, most kinematic distances being uncertain as they are, agree well with the RR Lyraes too, and the discrepant method cannot be identi ed unambiguously. In fact, with the SNe II scale being systematically smaller as well, it is the Cepheids which now appear discrepant (however, the SNe II method needs much further improvement to attain the same level of precision). Finally, we nd no evidence for contradiction between the four distance scales, in the sense, suggested by van den Bergh (1995), that a future correction of a systematic in the RR Lyrae distances could cause discrepancy for some, while removing it for others. This means that the RR Lyrae scale could be zeroed to the Cepheids without destroying the overall consistency between the four methods found here. A side e ect of this could be a decrease in the age estimates for globular clusters. As an outlook for the near future, we feel that all four methods have plenty of room for improvement { not one of them stands out as being of much superior accuracy to the rest. We expect that the validity of our combined approach to extragalactic distances will endure because the expanding data set will always remain fragmentary. Figure Captions Figure 1: Projections of the distance ellipsoids for four objects of our sample. The ellipses due to each method are marked, e.g. KD is for kinematic distance; the small unmarked ellipses are the projections of the 12-dimensional average distance ellipsoid. There are no classical Cepheids in globular clusters, therefore \wild guesses" were used (see text), hence Cepheid distance ellipses are very elongated or not seen at all in such cases. The panels are marked (a) to (d), clockwise, from the upper left one. Figure 2: Projections of the distance ellipsoids for all four methods for the two objects for which these are available { the Galactic Center and the Large Magellanic Cloud. The unmarked ellipse is the projection of the average distance ellipsoid. Note that while for the LMC the kinematics and Cepheids agree better than with the RR Lyrae, for the GC it is the Cepheids and RR Lyrae which agree better.	2019-04-14T01:49:27.017Z	1995-08-25T00:00:00.000	{ "year": 1995, "sha1": "4f40397265d1803bf4d4fb438f2cb0cb91f932a6", "oa_license": null, "oa_url": "http://arxiv.org/pdf/astro-ph/9508122", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "0e4768d7e00e9c76258e38f13503fca659129471", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
214023809	pes2o/s2orc	v3-fos-license	Periodic impulse signal separation based on resonance-based sparse signal decomposition and its application to the fault detection of rolling bearing The main purpose of the paper is to propose a new method to achieve separating periodic impulse signal among multi-component mixture signal and its application to the fault detection of rolling bearing. In general, as local defects occur in a rotating machinery, the vibration signal always consists of periodic impulse components along with other components such as harmonic component and noise; impulse component reflects the condition of rolling bearing. However, different components of multi-component mixture signal may approximately have same center frequency and bandwidth coincides with each other that is difficult to disentangle by linear frequency-based filtering. In order to solve this problem, the author introduces a proposed method based on resonance-based sparse signal decomposition integrated with empirical mode decomposition and demodulation that can separate the impulse component from the signal, according to the different Q-factors of impulse component and harmonic component. Simulation and application examples have proved the effectiveness of the method to achieve fault detection of rolling bearing and signal preprocessing. Introduction In general, when local defects occur in a rotating machinery, the corresponding vibration signal will comprise periodic impulse component along with other components such as rotating harmonic component and noise. The periodic impulse component usually presents condition of rolling bearing and the fault information is usually contained in the envelope part. Therefore, how to effectively extract the impact component and apply a proper envelope analysis on it is a fatal issue for fault detection. Considering the complexity of multi-component vibration signal and impulse component is weaker than the interfering component and bandwidth overlap with each other, the traditional signal separation method based on filter will reduce the amplitude of weaken impact component and introduce much noise consequently. It has been noted that resonance-based sparse signal decomposition (RSSD) acts as a new nonlinear signal analysis approach concentrating on signal resonance attribute instead of scale or frequency, as provided by the wavelet transform and Fourier transform. [1][2][3][4] This method decomposes multi-component signal into sustained oscillations and non-oscillatory transients; they are corresponding to high-resonance component and lowresonance component, respectively. Considering the particularity of rotating machinery, especially the research object of this paper is rolling bearing, the highresonance component corresponds to rotating harmonic component and noise, and low-resonance component corresponds to rolling bearing fault component consequently. As a result, the RSSD can be used to extract the weaken periodic impact component. It is apparent that the low-resonance component representing weaken periodic impact feature also contains noise, adopting effective measure to improve the signal-to-noise ratio (SNR) of low-resonance component is essential for enhancing the fault characteristic frequency resolution ratio of envelope spectrum. Several investigators have claimed 5-7 that empirical mode decomposition (EMD) is an self-adaptive signal processing. According to the non-stationary vibration signal characteristics of rotating machinery, EMD method has been successfully introduced into rotating machinery fault detection. [8][9][10][11][12][13][14] Using EMD method, it can decompose the low-resonance component into a number of intrinsic mode functions (IMFs), then determining principal IMFs that involve fault characteristic for demodulation analysis. Compared with Hilbert demodulating method, Teager energy operator has a faster and higher precision demodulation advantage. 15,16 Besides, energy operator can also be capable of increasing SNR of envelope spectrum and sharpening the spectral peaks which reveal the presence of fault feature. 17,18 On account of above advantages, in this paper, the principal IMFs of low-resonance component are analyzed by energy operator demodulating, and each one's instantaneous amplitude and instantaneous frequency can be calculated. Finally, according to the spectra of envelope spectrum, faults of bearing rolling can be detected. In all, the author proposed a novel hybrid intelligent fault detection that is based on RSSD combination with EMD and energy operator demodulating. By synthesizing and utilizing multiple advanced signal processing algorithm, this method not only overcomes the shortcomings of traditional signal separation method but also inherits the advantages of EMD and Teager energy operator. Finally, the framework of this method can be seen in the following. As local defects occurred in rolling bearing, RSSD can separate the impulse component from the vibration signal. Furthermore, considering the fault information exists in modulated amplitude, EMD method can be used as a pretreatment to decompose the impulse component into a number of IMF. Finally, using energy operator to analyze IMF component to get spectrum of instantaneous amplitude and rolling bearing fault can be detected consequently. As shown in Figure 1, the major section of proposed method is RSSD; the effect of decomposition is inseparably dependent on parameter value, such as Q-factor, redundancy parameter r, maximum decomposition level L. Value of these parameters can use optimization algorithm or manual setting. Signal resonance Generally, the resonance property of generic signal can be quantified by its quality factor, or Q-factor Q = f c =B W which is defined as the ratio of center frequency f c to its bandwidth B w . Furthermore, a higher Q-factor reflects signal with a more sustained oscillations in time domain and narrow-band attribute compared to lower Q-factor. As shown in Figure 2, it illustrates the concept of signal resonance attribute. Figure 1(a) and (b) essentially consists a single cycle waveform and they are all of low-resonance signal. It is noticeable that a low resonance can be either a high frequency signal or a low frequency signal and they can be converted to each other by time-scaled versions that infer time-scaled version does not affect its degree of resonance. As shown in Figure 2 the center frequency is approximately same and bandwidth coincides with each other are difficult to disentangle by linear frequency-based filtering, while they have difference Q-factors. Fortunately, RSSD can be able to separate them effectively. Tunable Q-factor wavelet transform RSSD utilizes tunable Q-factor wavelet transform (TQWT) [19][20][21] which achieves high Q-factor constant-Q (wavelet) transforms for the sparse representation of high-resonance component and low Q-factor constant-Q (wavelet) transforms for the sparse representation of low-resonance component with the aid of high Q-factor analysis and low Q-factor analysis, respectively. On the other hand, tunable Q-factor wavelet transformation which is developed in terms of iterated two-channel filter banks and can be implemented efficiently with fast Fourier transform (FFT) has a series of advantages, such as it is fully discrete has the perfect reconstruction property and can be modestly overcomplete. Twochannel filter banks for the TQWT that contain analysis and synthesis filter banks are shown in the following: symbol a represents low-pass scaling parameter, symbol b represents high-pass scaling parameter, b = 2=(Q + 1), a = 1 À b=r, symbol r represents redundancy parameter. The subband signal v 0 (n) has a sampling of af s ; likewise, the subband signal v 1 (n) has a sampling of bf s where f s is the sampling rate of input signal x(n). Finally, the signal decomposition based on TQWT is implemented by iteratively applying the twochannel filter bank on its low-pass channel. A L-stage wavelet transform is illustrated in Figures 3 and 4, W L is the wavelet high-pass subband signal produced by the L stage and C L is the wavelet low-pass subband signal produced by the L stage. Decomposition RSSD uses morphological component analysis (MCA) 22 to achieve nonlinear separation of vibration signal based on oscillatory property and establish the efficiency (sparsity) of high-resonance component representation and low-resonance component representation. Assuming an observed signal x = x 1 + x 2 , with x, x 1 , x 2 2 R N , the goal of MCA is to estimate x 1 and x 2 individually based on observed signal x. Providing that x 1 and x 2 can be sparsely represented in bases (or frames) S 1 and S 2 which are achieved by TQWT and S 1 and S 2 has low mutual coherence, respectively. An objective function of MCA can be represented in the following, and x 1 and x 2 can be estimated by minimizing this objective function With respect to W 1 and W 2 , MCA can provide the estimates c x 1 = S 1 W 1 and c x 2 = S 2 W 2 . Besides, as shown above, the minimization of objective function in Due to the non-differentiability and large number of variables in the objective function (equation (1)), RSSD uses split augmented Lagrangian shrinkage algorithm (SALSA) in combination with iteration to renew value of W 1 and W 2 to achieve the minimization of objective function. Finally, it can effectively achieve nonlinear separation of vibration signal into different components. Teager energy operator In general, a fault vibration signal contains multicomponent amplitude modulation-frequency modulation (AM-FM) that is shown in following As for each single-component x i (t), the corresponding Teager energy operator is defined as follows Using equations (4) and (5), the instantaneous amplitude and frequency of AM-FM component can be approximately computed Obviously, periodic impulse component reflects essential fault information; as a result, it is indispensable to compute instantaneous amplitude a i (t) of periodic impulse component by the way of Teager energy operator. Finally, using the envelope spectrum of instantaneous amplitude a i (t), it can detect fault feature frequency. Theoretical simulation In order to theoretically verify the effectiveness and superiority of proposed method, the author sets the following synthetic signal x(t) as shown in equation (6). In this synthetic signal, x 1 (t) is a periodic impulse component that represents rolling bearing fault, x 2 (t) is an AM component that represents harmonic interference and n(t) is a stochastic noise component whose ampli- In the above equations, symbol M represents impulse numbers in the periodic impulse component, A m symbolizes impulse amplitude, b represents attenuation coefficient, T symbolizes impulse interval which infers that fault characteristic frequency f c = 1=T, f re symbolizes resonance frequency, u(t) is an unit step function, f z is gear-mesh frequency and f r is rotating frequency. Finally, the simulated parameters are shown in Table 1. In this paper, setting sampling frequency of 1000 Hz and number of sampling points is 4000. Time domain waveform of synthetic signal and decomposition signal are shown in Figures 5 and 6, respectively; compared with harmonic interference and stochastic noise component, the amplitude of periodic impulse component is significantly weaker from the waveform. Figure 7 presents the power spectrum magnitude of multi-component signal; it can be seen that the power spectrum of harmonic component and periodic impulse component coincided with each other that are difficult to disentangle by linear frequency-based filtering. Considering the amplitude of periodic impulse component is weaker compared with interfering component, the demodulation effect by directly using energy operator demodulating is poor. Using the proposed method to decompose synthetic signal, according to the signal characteristics and practical experience, 1 decomposition parameter setting is as follows: high-resonance component Q 1 = 3, redundancy r 1 = 3, maximum decomposition level L 1 = 30, low-resonance component Q 2 = 1, redundancy r 2 = 3, maximum decomposition level L 2 = 10. Finally, the synthetic signal is decomposed into high-resonance component H(n), low-resonance component L(n) and residual component R(n), as shown in Figure 8. It is significant that the high-resonance component mainly reflects harmonic property and low-resonance component mainly reflects periodic impact property; as for residual component R(n), it has low energy which implies that the proposed method has a good signal reconstruction function. Yet, it is noticeable that the low-resonance component compared with the original periodic impulse component has weaker amplitude. This phenomenon occurs due to the fact that the relative values of parameters l 1 and l 2 mainly affect the energy distribution of decomposition result. On the other hand, the envelope spectrum of lowresonance component is shown in Figure 9; it is obvious that the envelope spectrum can significantly detect the fault characteristic frequency f c and its second harmonic generation (SHG). This phenomenon shows that the proposed method can effectively extract modulation information of periodic impact component. In order to analyze the superiority of the proposed method, the EMD integrated Teager energy operator is also applied to the simulating signal for comparison, and the envelope spectrum is shown in the following. The envelope demodulation spectrum is presented in Figure 10; it is prominent that the fault characteristic frequency is weaker and the spectrum presents noise component, and it infers that the proposed method has a better performance. Analysis of anti-noise Generally, an efficiently signal decomposition algorithm must have a good anti-noise performance. For performing anti-noise characteristic, adding stochastic noise with SNR of 0, 22, 24 and 28 dB, and the envelope spectrum of simulating signal with different SNRs is presented by proposed method as shown in Figure 11. As shown above, it is prominent that the fault characteristic frequency can be detected; however, when the SNR is low, the spectrum peak of fault characteristic frequency becomes weaker consequently. This phenomenon infers that it is necessary to adopt optimal algorithm for fault feature enhancement Experimental validation In order to further validate the effectiveness of this proposed method, experiments are conducted with SKF 6205-type rolling bearings, the parameters of test rolling bearings are shown in Table 2. The experimental rig is shown in Figure 12; the shaft can be driven using a motor which can be running at variable speed. Two SKF bearings are mounted for supporting the shaft, the one on the right side is healthy and the another one on the left side is seeded with inner raceway and outer raceway fault using electro-discharge machining. An accelerometer is installed on the top of faulty bearing to collect vibration signals. Finally, the fault signal is collected by DAQ card with LabVIEW; the signal processing algorithm uses MATLAB. The experiment is first conducted on a SKF 6205type bearing with an inner raceway fault, the fault diameter is 6.4 mm and fault depth is 3.4 mm, the experimental rotating frequency is 29.95 Hz, the sample frequency is 12,000 Hz; according to equation (9), the fault characteristic frequency of bearing for inner raceway fault is calculated to be 162.36 Hz. On the other hand, as for outer raceway fault, the fault diameter is 4.3 mm and fault depth is 3.6 mm, the experimental rotating frequency is 28.83 Hz, the sample frequency is 12,000 Hz; according to equation (10), the fault characteristic frequency of bearing for outer raceway fault is calculated to be 103.18 Hz Inner raceway fault experiment The signal of inner raceway fault is shown in Figure 13; it is significant that the experimental signal consists of impact component combination with much noise. Taking into consideration that the study object is rolling bearing in this paper, by the way of repeating test, the high Q-factor Q 1 and low Q-factor Q 2 are preliminarily sorted in four and two. As for the parameter r, its value has little effect on the decomposition result; in this paper, the corresponding highresonance redundancy parameter r 1 and lowresonance redundancy parameter r 2 are the same value three. Finally, as for the decomposition level L 1 and L 2 , the author takes into account the resolution of decomposition result and time consumption, the values are 30 and 12, and the signal decomposition is shown in Figure 14. As shown in Figure 14, the high-resonance component mainly reflects harmonic feature and noise and the low-resonance component presents impact characteristic. Finally, the envelope spectrum of first IMF component is presented, as shown in Figure 15. It can be seen from the figure that the fault characteristic frequency and rotational frequency exist in the spectrum; besides, the second harmonic frequency of fault component also exists in the spectrum. This phenomenon infers that the proposed method can effectively be used to detect fault. For comparison, the EMD integrated Teager energy operator is also used for analyzing the signal; the envelope demodulation spectrum is presented in Figure 16. The rotational frequency and fault characteristic frequency are shown in Figure 16. However, compared with the proposed method, the second fault characteristic frequency does not exist and the spectrum peaks is weaker. Figure 12. Test rig. Figure 13. Inner raceway fault signal. Outer raceway fault experiment As for the signal of outer raceway fault, the waveform is presented as shown in Figure 17. Compared with the waveform of inner raceway fault signal, the impact feature is more evident. Using the proposed method and the same decomposition parameters, the decomposition result is shown in Figure 18. As shown above, the high-resonance component reflects harmonic characteristic and noise, and the lowresonance component reflects impact feature that contains fault information (Figure 19). For comparison, the EMD-integrated Teager energy operator used is same for analyzing outer raceway fault signal; the envelope demodulation spectrum is presented in Figure 20. Compared with the proposed method, in spite of the fact that the above decomposition result can detect fault feature frequency f o and rotational frequency f r , the spectrum peak of fault characteristic frequency is weaker. It implies that the proposed method can improve the effectiveness of fault detection. Finally, the author considers that the reason why the proposed method can have better performance is that the RSSD can accurately extract the impulse component in vibration signal, and it is an essential signal preprocessing. This process can improve signal decomposition and fault detection. Conclusion In this paper, the author proposes a new method to separate the periodic impulse component among multicomponent signal and uses it to achieve rolling bearing fault detection. The following conclusions can be inferred from this paper: 1. RSSD can effectively separate impulse signal and harmonic signal based on Q-factor that can make up the disadvantages of linear frequencybased filtering; furthermore, compared with EMD and envelope analysis, the proposed method can be used for rolling bearing fault detection. 2. The proposed method in this paper can be appropriate to act as a signal preprocessing, the high-resonance component corresponds to harmonic characteristic and the low-resonance component can reflect the impact feature. Consequently, this process can identify fault characteristic more clearly. 3. The author discovers some parameters such as Q-factor has an evident effect on fault detection; in order to get better performance, the author must repeatedly modify the parameters. In the next step, the author considers that we can establish an optimization function for getting the optimal decomposition parameters. In this paper, the research object is impulse signal; therefore, it infers that kurtosis value of impulse signal can achieve this goal. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.	2020-01-23T09:07:31.328Z	2020-01-16T00:00:00.000	{ "year": 2020, "sha1": "bf1f4408c2944bc793b7a19fe6e3b0867079e015", "oa_license": "CCBY", "oa_url": "https://journals.sagepub.com/doi/pdf/10.1177/0020294019866858", "oa_status": "GOLD", "pdf_src": "Sage", "pdf_hash": "70f48329406bc8eab9fe98993201fb78cc361ff7", "s2fieldsofstudy": [ "Engineering", "Computer Science" ], "extfieldsofstudy": [ "Physics" ] }
235462453	pes2o/s2orc	v3-fos-license	Functionalized MWCNTs-quartzite nanocomposite coated with Dacryodes edulis stem bark extract for the attenuation of hexavalent chromium Multiwalled carbon nanotubes/quartzite nanocomposite modified with the extract of Dacryodes edulis leaves was synthesized and designated as Q, which was applied for the removal of Cr(VI) from water. The adsorbents (PQ and Q) were characterized using the SEM, EDX, FTIR, TGA, XRD, and BET analyses. The XRD revealed the crystalline composition of the nanocomposite while the TGA indicated the incorporated extract as the primary component that degraded with an increase in temperature. The implication of the modifier was noticed to enhance the adsorption capacity of Q for Cr(VI) by the introduction of chemical functional groups. Optimum Cr(VI) removal was noticed at a pH of 2.0, adsorbent dose (50 mg), initial concentration (100 mg dm−3), and contact time (180 min). The kinetic adsorption data for both adsorbents was noticed to fit well to the pseudo-second-order model. The adsorption equilibrium data were best described by the Langmuir model. The uptake of Cr(VI) onto PQ and Q was feasible, endothermic (ΔH: PQ = 1.194 kJ mol−1 and Q = 34.64 kJ mol−1) and entropy-driven (ΔS : PQ = 64.89 J K−1 mol−1 and q = 189.7 J K−1 mol−1). Hence, the nanocomposite demonstrated potential for robust capacity to trap Cr(VI) from aqueous solution. Determination of pH point of zero charge (pH PZC ). To determine the pH at the point of zero charges of PQ and Q, about 0.1 g of the materials was transferred into eleven 250 cm 3 glass Erlenmeyer flasks with each containing 50 cm 3 of 0.1 mol dm −3 NaCl solution at pH values ranging from 2 to 12. The flasks were stoppered and agitated for 48 h in a preset thermo-regulated water bath at 25 °C. The final pH of the mixture was obtained after 48 h and a plot of the final pH versus the initial pH was made from which the pH PZC of PQ and Q was extrapolated from the line intercept 55 . Batch adsorption experiments. Batch To assess the thermodynamics of the adsorptive process, the initial concentration experiment was repeated for 303 K, 308 K and 313 K. The residual concentration of Cr(VI) was estimated by the colorimetric method using 1,5-diphenyl-carbazide as a complexing agent, analyzed using the UV-visible spectrophotometer at 540 nm 56 . The adsorption capacities and the uptake efficiency of PQ and Q were calculated as shown in the supplementary information. Kinetics and isotherm models. The kinetic modeling of adsorption was performed using four kinetic models while the isotherm analysis was carried out by applying eight isotherm models as described in the supplementary information. Reusability experiments. To examine the reusability of PQ and Q, the adsorbents were used to remove the adsorbate (Cr(VI)) from the aqueous phase using the same adsorption procedure stated in the previous section. The regeneration of PQ-Cr and Q-Cr was performed by making use of NaOH. About 0.5 g of PQ-Cr or Q-Cr was in contact with 25 cm 3 of 0.5 mol dm −3 NaOH for 3 h at 25 °C. The regenerated adsorbents were washed twice with ultrapure water and dried for the next cycle. The removal efficiency of PQ and Q for the next cycle was estimated using Eq. (2). Compliance with ethical standard. The collection of plant materials complied with relevant institutional, national, international guidelines and legislations. Results and discussion Characterizations of adsorbents. The micrographs of the surface morphology of PQ and Q are displayed in Fig. 1. The SEM micrograph of the pristine quartzite (PQ) revealed an aggregate of smooth particles with varied shapes and sizes (Fig. 1a). The structure of PQ reflected the metamorphic nature of quartzite rock. The micrograph (Fig. 1b) of the nanocomposite(Q) showed that the quartzite particles were rapped with an intertwined network of cylindrical tubes resulting in the formation of microscopic channels on the surface with an increase in surface roughness probably due to coating by the incorporated extract. The micrograph (Fig. 1c) of the nanocomposite adsorbed with Cr(VI) (Q-Cr) was acquired at a higher magnification which further revealed an increase in surface roughness of the grains without the segregation of the network of carbon nanotubes or comparable sign of degradation concerning the surface integrity of the nanocomposite. This indicated the stability of the material in the aqueous environment. To further confirm the adsorption of chromium onto the surface of Q, the acquired EDX of Q-Cr (Fig. 1d) revealed the surface adsorbed chromium. It should be noted that the percentage (0.47%) indicated by EDX represented Cr on the spot being scanned and not the total adsorbed Cr on the nanocomposite. X-ray diffraction analysis was used to assess the mineral content of quartzite. As shown in Fig (Fig. 1d) showed consistency with the identified minerals in the diffractogram. The diffractogram of the nanocomposite (Q) identified the same mineral with a reduction in the intensities of the peaks which could be associated with the surface coating and the presence of the carbon nanotubes. However, the diffraction patterns associated with the carbon nanotubes were not observed, which indicated the dominance of the quartzite mineral intensities due to higher crystallinity. www.nature.com/scientificreports/ The FTIR spectra of PQ and Q were recorded in the wavenumber range of 4000-400 cm −1 . The prominent peaks in the spectrum of PQ (Fig. 3) appeared in the range 1000-400 cm −1 . These peaks were associated with quartz, which indicated Si-O-Si 60 . However, additional bands were observed in the spectrum of the nanocomposite at (ν/cm −1 ): 1380, 1610, 3500 assigned to asymmetric vibrational stretching modes of carboxylate group (-COO-), the symmetrical C=O stretching of ionic carboxylate groups and the -OH stretching [61][62] . These peaks indicated the incorporation of the Dacryodes edulis leaves extract as well as the functionalized carbon nanotubes on the surface of the adsorbent. In a bid to assess the interaction of the adsorbate on the surface of the adsorbents, the FTIR spectra of the loaded adsorbents (PQ-Cr and Q-Cr) were acquired. As shown in Fig. 3, the intensity associated with the band of the hydroxyl functional group on the surface of the nanocomposite was reduced after the adsorption step. This suggested that the -OH functional groups were actively involved in the adsorption of Cr(VI) onto the nanocomposite. The surface area and pore volume of PQ and Q were assessed using the BET nitrogen adsorption-desorption technique. Meanwhile, the pore diameter of PQ and Q were estimated using the Barrett-Joyner-Halenda (BJH) approach. The physisorption isotherm of PQ indicates poor adsorbate-adsorbent interaction. However, the incorporation of carbon nanotubes and plant extract from Dacryodes edulis showed a distinct improvement in the specific surface area of the nanocomposite over PQ (Fig. 4). This indicated an increase in the rate of interaction due to the availability of more surface area. Similarly, the increase in pore diameter and pore volume (Table 1) of the nanocomposite (Q) compared to PQ indicated the introduction of macropores along with the existing mesoporous structure of PQ. The form of the graph of amount gas adsorbed versus partial pressure is known as an adsorption isotherm, which are classified as type I-V according to their shape 63 . A type III physisorption isotherm was observed for PQ while a type IV isotherm was observed for the nanocomposite, which indicated the enhanced porosity of the nanocomposite. The thermal stability of Q was investigated using the thermogravimetric analysis technique. The thermal behavior of pristine quartzite as reported by Xing et al.revealed the loss of physisorbed water only 64 . However, the presence of plant extract and carbon nanotubes strongly influence the thermal stability of the nanocomposite. Figure 5, displayed the thermogram of the nanocomposite which revealed a unique decomposition pattern. www.nature.com/scientificreports/ About 1.90% mass loss was noticed with the rise of temperature to 110 °C. This could be attributed to the loss of physisorbed water. A further reduction of 2.06% observed as the temperature increased to 300 °C could be associated with the loss of decomposed plant extract on the surface of the nanocomposite and the loss of internally bonded water. However, a progressive decline in mass was observed in the temperature range of 300-800 °C, which reflects the elimination of volatile inorganic or organic materials that were internally bonded. Over the investigated temperature range, about 17.42% mass loss was noticed for the nanocomposite. This accounted for the incorporated plant extract and it showed the stability of quartzite framework, which provided support for the carbon nanotubes used in the fabrication of the nanocomposite. Hence, the fabricated adsorbent can be used to treat wastewater even at high temperatures (< 300 °C) without loss of morphological integrity. Effect of pH on Cr(VI) uptake. Figure 6 revealed the implication of varying sorbate pH on the adsorptive removal of Cr(VI) by PQ and Q. The optimum removal capacities of PQ (27.50 mg g −1 ) and Q (45.87 mg g −1 ) were observed at pH 2. It is noted that at all pH values, the surface modification of PQ to give Q, led to about twice as much removal capacity of Q than for PQ. Meanwhile, the adsorptive capacity of PQ and Q was noticed to decrease with decreased acidity of Cr(VI) solution. Hence, the nanocomposite (Q) demonstrated an adequate potential to eliminate hazardous hexavalent chromium from the aquatic ecosystem. Speciation of hexavalent chromium and the pH PZC of the adsorbents offers better insight on the high Cr(VI) uptake capacity of the sorbent at low solution pH. Note that a reducing argent is necessary to reduce Cr(VI) to Cr(III) at a low pH. As shown in Fig. 7, the pH PZC of PQ and Q were 4.08 and 6.18, respectively. This indicates that at solution pH below and above these values, the surface of the adsorbents (PQ and Q) will be positively and negatively charged respectively. Hence, at solution pH 2, the adsorbents will be positively charged. The hexavalent chromium exists in different oxyanions form with variation in solution pH. In the pH range of 2 to 4, Cr(VI) exist mainly as HCrO 4 in an aqueous solution. Meanwhile, as the pH increase from 4 to 6, HCrO 4 − is converted to and in equilibrium with Cr 2 O 7 2- [65][66] . The enhanced uptake capacity of PQ and Q at pH 2 could be due to electrostatic interaction between the positively charged surface of the adsorbents and negatively charged chromium species. It is also possible that Q acts as a reducing agent during the Cr(VI) uptake process, due to the reducing potential of Dacryodes edulis 54 . However, finding from this study is in good agreement with the reports from other authors that got optimum pH of 2 22,67 . www.nature.com/scientificreports/ Fig. 8, the time-dependent adsorptive removal of Cr(VI) by PQ and Q was in three stages. The first stage was a fast phase (occurred before 20 min), followed by a gradual increase until 180 min and the last stage that involved no significant increase in the uptake capacity of PQ and Q. Hence, 180 min was selected as the optimum contact time for the elimination Cr(VI) from aqueous solution. However, to ensure complete removal of the adsorbate, 1440 min was employed for further experiment. PQ adsorbed more than double the amount that Q adsorbed at all times > 20 min. Kinetics study. To shed light on the Cr(VI) removal rate by PQ and Q, kinetics model namely pseudo-firstorder, pseudo-second-order, Elovich, and Moris-Weber intraparticle diffusion were employed. The nonlinear equations of these models were used to analyze the kinetic data (see Table 2) and the acquired plots are shown in Fig. 9. Meanwhile, the least sum of square residuals (SSR) was used as the goodness-of-fit measure in selecting the model that best describes the data ( Table 2). The model with the smallest SSR value gave the best fit to www.nature.com/scientificreports/ the experimental. The pseudo-first-order kinetic model is best used to describe the adsorption process that is dominantly controlled by diffusion. This model is also based on the fact that an adsorbate binds to a single adsorption site. On the other hand, an adsorption process that is driven by chemisorption is best described by pseudo-second-order kinetic models. In principle, the pseudo-second-order kinetic model assumes binary adsorption of a sorbate. An adsorptive model that demonstrates chemisorption as its rate-determining step is described by the Elovich kinetic model. However, Morris-Weber intraparticle diffusion is used to describe the basic stages involved in the adsorption process. These stages include (i) conveyance of sorbate from the solution bulk to a thin film layer, (ii) transport of sorbate from film layer to the surface of the adsorbent, (iii) migration of sorbate from the adsorbent surface to the interior of the porous structure and (iv) adsorption of sorbate to the adsorption sites. As shown in Table 2, the adsorptive removal of Cr(VI) by PQ and Q was best described by pseudo-second-order (SSR = 10.87) and Elovich kinetic models (SSR = 32.32) respectively. This suggested that chemisorption was the rate-limiting step in the adsorptive removal of hexavalent chromium from the aquatic solution. Effect of adsorbent dose. As shown in Fig. 10, an increase in removal efficiencies of PQ (15.4 to 67.7%) and Q (73.1 to 98.7%) was noticed as the adsorbent dose was increased from 0.01 to 0.4 g (see Fig. 10). This could be attributed to the increasing amount of adsorption sites at fixed initial Cr(VI) concentration. On the contrary, the uptake capacity of PQ (38.5 to 4.2 mg g −1 ) and Q (182.7 to 6.2 mg g −1 ) were observed to decrease with increased dosage. The decrease in the uptake capacity could be associated with the agglomeration of the sorbent at a higher dosage. Irrespective the adsorbent dose, Q outperformed PQ by having about double the removal capacity of PQ. Effect of initial concentration and solution temperature. The implication of initial concentration on the adsorptive removal of Cr(VI) by PQ and Q was investigated and the results are displayed in Fig. 11. This www.nature.com/scientificreports/ study was examined over a concentration range of 10-100 mg g −1 with a fixed adsorbent dose of 0.25 g. As shown in Fig. 11, the result revealed that the uptake capacities of PQ and Q increased from 2.5 to 24.3 mg g −1 and 5.1 to 44.0 mg g −1 respectively as the initial Cr(VI) concentration increased from 10 to 100 mg dm −3 at 298 K. It indicated that the adsorptive removal of Cr(VI) onto PQ and Q was strappingly dependent on the initial hexavalent chromium concentration 68 . The phenomenon could be associated with enhanced collisions frequency between the Cr(VI) ions and the adsorption sites on the surface PQ and Q, resulting in higher surface coverage at high initial Cr(VI) concentrations, and thus high uptake capacities. A similar trend was observed at all temperatures investigated. However, an increase in solution temperature was noticed to enhance the uptake capacity of Q, although this effect was more significant at higher solution temperature and thus, demonstration the endothermic adsorptive process. In contrast to this, the influence of solution temperature on the adsorption of hexavalent chromium onto the surface of PQ was trivial. The quantity of Cr(VI) adsorbed at equilibrium, q e , is ca double for PQ compared to the Q at the same initial concentration of Cr(VI). Adsorption isotherm. The distribution of sorbate in the solid-liquid interphase and the estimation of sorbent potential to eliminate specific sorbate can be determined by making use of mathematical models termed adsorption isotherm. To understand the mechanism of Cr(VI) adsorption onto the surface of PQ and Q, various two-and three-parameter isotherms were used to analyze the equilibrium adsorption data. The least sum of squared residuals (SSR) and the residual squared errors (RSE) were used to select the model that best describes the experimental data. A smaller sum of squared residuals value indicate a better fit of the model to the data; a value of zero means a perfect fit of the model. Among the two-parameter models (Dubinin-Radushkevick, Temkin, Freundlich, Langmuir), the Langmuir isotherm was adequate to describe the adsorption of Cr(VI) onto PQ and Q within the studied temperature range observed (see Table 3). This suggested that the adsorption process of Cr(VI) by PQ and Q was mainly monolayer adsorption. However, those of the three-parameter models (Toth, Redlich-Peterson, Khan, Sips), Sips and R-P were noticed to best fit PQ and Q respectively. Comparison of adsorbents for Cr(VI). A comparison of the maximum monolayer capacity (q max ) of PQ and Q, with previously reported adsorbents, showed that the nanocomposite reported in this study possess a better absorbance capacity for application in environmental remediation practice (see Table 4). Especially is noted that the surface modification of PQ led to about five times the maximum removal capacity for Q, compared to PQ. Adsorption thermodynamics. Evaluation of the thermodynamics of adsorption was performed as decribed in the supplementary information. Table 5 revealed that at all temperatures investigated, the ∆G° values of the adsorption of Cr(VI) onto PQ and Q were negative. It showed that the uptake of Cr(VI) by PQ and Q were feasible and spontaneous. Meanwhile, a slight increase in the ∆G° values was noticed with increased solution temperature. This indicated that the removal efficiency of the adsorbents was favored at higher solution temperatures. Furthermore, positive values of ΔH° and ΔS° were estimated from the thermodynamic analysis. This indicated an endothermic adsorption process and an increased haphazardness at the sorbate-sorbent interface respectively. It is worth mentioning that the thermodynamic of an adsorption process creates a path to unveiling the adsorptive mechanism. However, literature revealed that the adsorption process with ΔH° values between 2.1 and 20.9 kJ mol −1 is physisorption [77][78] , while ΔH° values are between 80 to 200 kJ mol −1 is chemisorption driven 77 . Hence, regarding the ΔH° values of PQ and Q as display in Table 4, the adsorptive removal of Cr(VI) by PQ was a physisorption process while the uptake of Cr(VI) onto Q was a physicochemical process 1 . Reusability of PQ and Q. The tendency of PQ and Q to retain their adsorption efficiency after multiple usages were examined and displayed in Fig. 12. To achieve this, the adsorption-desorption cycle was repeated five times. After the fifth cycle, PQ and Q were noticed to have an efficiency of 29% and 78%. The high reusability of Q after 5 cycles is an indicator of the excellent design of Q for Cr(VI) absorbance from waste water. This Figure 11. The effect of initial concentration on uptake of Cr(VI) by PQ and Q. Adsorption mechanism. The mechanism responsible for the adsorption of Cr(VI) uptake onto Q may be due to the surface modification of MWCNTs-quartzite using Dacryodes edulis leaves extract. The optimum uptake of Cr(VI) onto PQ and Q was established at pH2 (see Fig. 6). Meanwhile, the pH PZC of PQ and Q were 4.08 and 6.18 respectively (see Fig. 7). Hence, at pH2 the positively charged surface of PQ and Q would interact with oxyanions species of Cr(VI) electrostatically. The functional groups of the phytochemical constituent of most plant extract, confer reductive characteristics on extract obtained from different part of plant. [REF]. Hence, the reduction of Cr(VI) to Cr(III) followed by the adsorption of Cr(III) onto the surface of Q via electrostatic interaction may be a possible route for the uptake of Cr(VI). The uptake of Cr(III) could also be via pore entrapment (the enhanced pore volume, pore dimeter and surface area of the nanocomposite (Q) could aid pore entrapment, see Table 1). As shown in Fig. 3, the FTIR spectra of the spent adsorbent (Q-Cr) revealed the disappearance of -OH bands, demonstrating that Cr(VI) chemically interacted with the surface of Q via chemisorption. The pseudo second order model was observed to best describe the uptake of Cr(VI) on to Q (see Table 2), This further justified the inclusion of chemisorption in the uptake of Cr(VI) onto Q. Conclusion A newly fabricated nanocomposite adsorbent (Q) prepared from multiwalled carbon nanotubes (MWCNT) and quartzite (PQ) coated with plant extract was assessed for its capacity to trap/reduce Cr(VI) to its nontoxic species from an aqueous solution. The nanocomposite (Q) showed good surface morphology, undisrupted crystalline phases, enhanced thermal stability, improved pore dimeter and enhanced surface area. Meanwhile, the optimum adsorptive conditions for removal of chromium(VI) by Q was established to be pH 2, adsorbent dose of 50 mg, initial chromium(VI) concentration of 100 mg dm −3 , the temperature of 318 K, and a contact time of 180 min. Under this conditions Q gave a maximum monolayer capacity (q max ) of 192.5 mg g −1 . Furthermore, the experimental isotherms data obtained for the uptake of chromium(VI) onto PQ and Q were best described by the Langmuir model. The time-dependent adsorption data were best described by pseudo-second-order kinetic. The estimated thermodynamic parameters suggested that the removal of chromium(VI) by PQ and Q was spontaneous, endothermic, and entropy-driven. The adsorbent (Q) has demonstrated robust efficiency and large absorbance capacity for the removal of chromium(VI) from aqueous solutions. Owing to the excellent reusability of the adsorbent, Q, giving 78% Cr(VI) absorbance after 5 cycles, Q are recommended to be tested scaled-up conditions for industrial applications.	2021-06-18T06:16:23.812Z	2021-06-16T00:00:00.000	{ "year": 2021, "sha1": "7f222ecaa8f4a327b8de646eee0cf2702277f50b", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41598-021-92266-0.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "355e1b6e2430401bd8b0b5dd7749702b3db07ca2", "s2fieldsofstudy": [ "Engineering" ], "extfieldsofstudy": [ "Medicine" ] }
234233052	pes2o/s2orc	v3-fos-license	PeachNet: Peach Diseases Detection for Automatic Harvesting To meet the food requirements of the seven billion people on Earth, multiple advancements in agriculture and industry have been made. The main threat to food items is from diseases and pests which affect the quality and quantity of food. Different scientific mechanisms have been developed to protect plants and fruits from pests and diseases and to increase the quantity and quality of food. Still these mechanisms require manual efforts and human expertise to diagnose diseases. In the current decade Artificial Intelligence is used to automate different processes, including agricultural processes, such as automatic harvesting. Machine Learning techniques are becoming popular to process images and identify different objects. We can use Machine Learning algorithms for disease identification in plants for automaticharvesting that can help us to increase the quantity of the food produced and reduce crop losses. In this paper, we develop a novel Convolutional Neural Network (CNN) model that can detect diseases in peach plants and fruits. The proposed method can also locate the region of disease and help farmers to find appropriate treatments to protect peach crops. For the detection of diseases in Peaches VGG-19 architecture is utilized. For the localization of disease regions Mask R-CNN is utilized. The proposed technique is evaluated using different techniques and has demonstrated 94% accuracy. We hope that the system can help farmers to increase peach production to meet food demands. Introduction The food demands of approximately 7 billion people on Earth are met using modern technologies in agriculture and related industries. There are certain factors, such as climate change, pollinator decline, and plant diseases, that threaten the security of food production. It has been reported that approximately 50% of yields are lost because of pests and diseases in plants. These diseases affect crop production quality and quantity. Therefore, it is important that we are able to diagnose these diseases at an early stage to improve the quality of crops and minimize production losses. For inexperienced farmers, it is difficult to identify diseases in plants through observation with the naked eye. Even if they are able to identify such concerns, it may not be possible for them to find the appropriate treatment to cure the disease. At the same time, detecting diseases by hiring expert farmers to give their opinions or use traditional tools, and even investigating plants in laboratories, can be time consuming and prone to errors. Novel and quick techniques that can detect pests or diseases in crops are needed to enable us to develop control measurement techniques with higher efficiency. Researchers have developed different scientific mechanisms to control the loss of crop yields due to diseases. A variety of pesticides are used to control these diseases in plants or food. It is an important scientific contribution to correctly identify diseases in plants and use appropriate pesticides. Traditionally, these diseases have been identified by plant experts or local clinics specifically developed for plant disease control. Due to the widespread use of the Internet, this expertise has become available online. More recently, mobiles or robots have been developed to identify diseases in plants. The development of an accurate system that can detect diseases in plants and recommend a diagnosis is a crucial and important step to developing an automated harvesting mechanism. This is important to protect plants and fruits from diseases and to increase the rate of production. If diseases are not detected accurately, appropriate treatment cannot be recommended and, therefore, the quality and quantity of food items cannot be increased. However, this automatic detection of diseases in plants is challenging due to multiple factors, such as variation in illumination, occlusion of plants or fruits, and the fact that fruits have similar appearances. There is a need to develop an automatic and intelligent system that can accurately detect diseases, identify the location of the disease, and recommend pesticides to control the spread. This will help the agriculture industry to produce quality items and increase the quantity to meet food demands. Recently, tremendous progress has been shown in computer vision and object detection and recognition [1]. Different Convolutional Neural Network (CNN) models have been developed and have surpassed human performance. Although these models are time consuming to train on large datasets, once they are trained, they can efficiently detect objects and make classifications and hence can be used with consumer devices, such as smartphones or robots. Techniques based on Deep Learning (DL) algorithms have been demonstrated successfully for the development of endto-end applications. This means that input data are processed and classified into different classes, without involving any hand-engineered features extraction. These algorithms work as a mapper to map input data to output labels using stacked layers in CNN. The overall objective is to fine-tune the parameters in the training of these models. The detection of diseases in plants using imageprocessing techniques is a landmark contribution in computer vision [2]. Most farmers do not have the training to correctly guess the disease affecting their plants and the appropriate treatment to apply, therefore, automatic disease detection in plants can help them to detect disease and recommend appropriate treatment. This is an important contribution in the field of automatic harvesting, as robotic machines can look after the field and identify diseases in plants or fruits and take the necessary actions to control the spread of disease to improve the quality of food and increase production. To meet food requirements, the United Nations Food and Agriculture Organization (FAO) has estimated that by 2050 agriculture needs to be increased by 70%. They have also observed that the usage of chemicals as pesticides has negative effects on agro-ecosystems. According to the FAO, peach trees are threatened by diseases affecting their growth and quality, such as Bacterial Canker, Bacterial Spot, Crown Gall, Peach Scab, and others. It is reported that only 16% of peaches produced are globally traded, demonstrating the importance of peaches in domestic markets and the security of the fruit. Current cutting-edge research, such as Artificial Intelligence (AI), robotics, Internet of Things (IoT), big data, cloud computing, and Machine Learning (ML) will aid in the transfer of agriculture and support farmers in the future. In this paper, we develop an intelligent system that detects diseases in peach trees and fruits using Deep Convolutional Neural Networks (DCNN), which can be generalized to multiple tasks subject to different environmental conditions, such as brightness, occlusion, and background similarity. Previous studies have either created their own dataset of images or used standard datasets, such as IPM Images, 1 PlantVillageImages [3], Fruit-360 [4], and the APS Image database. 2 For the evaluation of the proposed work, we have demonstrated results that we compare to other research in the field, both qualitatively and quantitatively. To the best of our knowledge, this is the first system that can accurately and efficiently detect diseases in peach trees and localize to the region of disease. Different standard evaluation techniques are used, such as precision, recall, f1-score, ROC curves, and a confusion matrix. The dataset, along with the labelling of images, will be shared with the community through the online publication of the research to support researchers in extending these findings. The main contributions of this research are: • Developing an efficient and intelligent disease-detection system in peaches that is trained rapidly and has a limited set of images; the system uses a CNN that is trained on a larger set of images known as ImageNet [5]. • Showing our results to the scientific community by sharing our dataset, results, and documentation. • Improving the accuracy of peach disease diagnosis. • Facilitating the work of peach farmers with an AI-based peach-disease and pest-detection system based on CNN. The paper is arranged as follows. We provide a detailed literature review in Section 2. The proposed methodology of disease detection in peaches is given in Section 3. The results collected in our experiments are demonstrated in Section 4. The paper concludes and discusses future directions of the research in Section 5. Literature Review A smartphone-assisted disease identification technique using DL was developed in [6]. A dataset of 54,306 images containing healthy leaves and diseased leaves were collected in a controlled environment. The developed technique identified 14 crop species and 26 types of crop diseases, and the proposed model achieved an accuracy rate of 99.35%. DeepFruits [7] is an accurate, efficient and reliable system developed to identify fruits in fields for automatic harvesting. The proponents developed a CNN that can detect different types of fruits and used a pre-trained VGG16 network based on ImageNet. Red, Green, and Blue (RGB) and Near Infrared (NIR) are used for early and late fusion and have demonstrated success over a single DCNN. They have achieved precision and recall of 0.807 and 0.838, respectively. In addition to the accuracy, the system can be developed quickly to deploy for different fruits. In [8], a deep neural network-based system was developed to identify plant diseases by analyzing leaf images. The model is able to identify 13 types of plant diseases by analyzing leaves, in addition to distinguishing leaves from their surroundings. They have demonstrated an accuracy rate of 96.3% in different plant diseases. In [9], an efficient DL model was developed to detect diseases in olive trees. The authors used the transfer learning approach along with data augmentation techniques to produce images of a balanced number of each category and to enable the model to work in different complex environments where new datasets are used. They have demonstrated higher accuracy, precision, recall, and f1-measure. In [10], a plant disease diagnosis for smartphone applications was developed. The authors extracted classification features with additional information, such as weather conditions, and developed disease signatures. These signatures were created by applying a statistical process to a small number of images used in training. Features extracted from images in the test dataset were compared with disease signatures to determine the highest-probability disease. This classification is independent from scale, orientation, or resolution of images. Accuracy of the system was reported as 70%-99%. In [11], an AI-based system was developed that can detect disease and pests in bananas. The authors developed a large dataset where images of diseases and pest symptoms in bananas were preprocessed by experts in the field. Three CNN architectures were trained using the transfer-learning technique. Six CNN models were developed for 18 diseases in banana plants. The study revealed that ResNet50 and InceptionV2 demonstrated the highest performance, with an accuracy of 90%. In [12], CNN architecture was trained to detect diseases in maize and peaches. An accuracy of 99.28% was achieved. In [13], multiple CNNs were combined to detect disease in tomato plants. They used a combination of Faster R-CNN and Mask R-CNN. To distinguish between healthy and diseased tomato plants, Faster R-CNN was used, and Mask R-CNN was applied to categorize 11 disease types and identify the location of the disease. In [14], a novel technique was used that can detect diseases in crops without limiting the approach to a specific crop. It was demonstrated to be much improved over the classical disease detection approach. The study examined whether fine tuning a pre-trained model on plant identification is more effective than a general object-recognition task. The findings demonstrated the visualization of features that were learned during the training. In [15], a review was provided that explains DL techniques for visualization of diseases in plants. The discussion described papers that have implemented different techniques of visualization that can detect and classify symptoms of diseases in plants. In [16], the effectiveness of different DL approaches for plant disease detection was evaluated. Transfer-learning and feature-extraction methods were used. The evaluation demonstrated that deep-features extraction produced better results than did transfer learning. They also demonstrated that VGG16 and VGG19 produced better results compared to other famous CNN models. In all previous studies, different DL techniques were used to detect diseases or identify different plants. Some studies have developed a combination of CNNs and have achieved better accuracy. The current study uses a combination of VGG-19 and Mask R-CNN to identify diseases in peach trees and fruit. To the best of our knowledge, this research is the first study that has extensively examined an AI-based system for automatic harvesting of peaches. This is the first step in automatic harvesting and will help us to increase the quality and quantity of peach production. Materials and Methods For classification purposes, the dataset used is important. We need to collect and process as much data as possible. Most available datasets, such as PlantVillage or Fruit-360, contain images of fruits and diseases, but these images are collected in a controlled environment where the object is clearly visible and there is appropriate lighting. However, in real-world situations, we do not have such a setting. Sometimes fruits are occluded, or there is not enough light to see clearly. Therefore, we have developed our own dataset, named "Peach-Disease," where we have taken images from existing datasets along with images that we collected in an uncontrolled environment. In the latter setting, we collected images under direct sunlight, under a canopy, on a cloudy day, in the shade, and in other similar conditions. Various peach experts visited several peach farms to collect images of various peach diseases. This combination of images (see Fig. 1) will help us to develop a robust model. Figure 1: A montage of images in the Peach-Disease dataset We used the augmentation technique to further increase the quantity of images. The main objective of augmentation is quantitatively increasing images inthe dataset and introducing some distortion in the images that will help the model to reduce overfitting during the training. Overfitting occurs when the model describes random error or noise other than the relationships in the data. Augmentation involves different transformation techniques, such as affine transformation, perspective transformation, and image rotation. The LabelImg software was used to perform the tagging process. The format used by ImageNet (PASCAL VOC) was adopted in this dataset to store the coordinates and labels of the boxes in an XML file. Depending on the number of disease-infected regions of the plants, each image may contain more than one annotation. We collected 3,199 images of healthy and diseased peaches. Of these, 2,782 images were collected from open datasets, and 417 images were collected in uncontrolled environments and labelled by peach experts. To process the images in the model proposed in this paper, we resized the images to 244 × 244 pixels. A montage of images collected in the Peach-Disease dataset is shown in Fig. 1. Peach experts have classified images of 11 common diseases that can occur in peach trees or fruits. The names of these diseases and the corresponding sample images are given in Tab The framework of the system PeachNet that can detect diseases in peach trees or fruits is given in Fig. 2. After collecting the images, we labelled the images of diseases and localized the diseased regions. Then we split the data into train, validation, and testing datasets, with 80% in training, 10% in validation, and 10% in testing. Then, we performed object detection using pretrained VGG-19 [17] architecture to identify the Region of Interest (RoI) in the peaches. Next, we used Mask R-CNN [18] to segment disease regions in RoI. We performed different evaluation metrics, such as precision, recall, f1-score, ROC, and a confusion matrix. We also compared the performance of the proposed technique with other state-of-the-art approaches. After training, the model can be deployed for use in automatic harvesting. Many recent advancements have been made in the domain of deep CNN for classifying images and performing object detection. However, it is still challenging to accurately detect objects in different machine-learning and computer-vision domains. Object detection and localization not only detect objects in a given image but also locate the region of the object in a given image. The VGG network architecture was developed at the University of Oxford. In this process, 3 × 3 convolutional filters are stacked in increasing depth, and volume size is reduced by max pooling. A stack of five convolutional layers where each layer is followed by a max-pooling layer is used. The final max-pooling layer is followed by three Fully Connected (FC) layers and a softmax layer. The two FC layers have 64 × 64 (4,096) channels. The last channel has 13 channels to detect different types of diseases in peach trees. In VGG-19, there are 19 weight layers in the network. The input layer received an input image of size 244 × 244. A sample VGG-19 architecture that is used for bounding box prediction in PeachNet is given in Fig. 3. In Faster R-CNN, one branch is used to classify and make a bounding box on RoI. This framework is extended in Mask R-CNN with another branch that can predict object masks in parallel to the branch used by Faster R-CNN. The bounding box of candidate classes in Faster R-CNN is detected by a Regional Proposal Network (RPN). These regional proposals are processed by RoIPOOL to extract features and perform classification and regression of the bounding box. Alongside the operation performed by Faster R-CNN, Mask R-CNN outputs a binary mask for each RoI. If the Intersection over Union (IoU) of a given RoI is at least 0.5, then it is considered positive, otherwise, it is considered negative. The positive RoIs are used to define the loss in the mask. Non-max suppression is performed after the box prediction. The highest-scoring box is processed by the mask branch. ResNets of depth 50 and 101 layers are used as the convolutional backbone architecture for feature extraction on the entire image. A sample architecture of Mask R-CNN used in PeachNet for segmentation is given in Fig. 4. Experimental Results The configuration of hardware and software used for the experiment performed here is shown in Tab. 3. The 10-fold cross-validation technique is used for evaluating the performance of the trained model. The mean Average Precision (mAP) is utilized calculate the accuracy of peachdisease detection. The average of the true positive class and the true positive plus false positive classes are calculated to compute the mAP, as shown in Eq. (1). The mAP for PeachNet is computed as 94%. The accuracy and loss computed for each iteration in the training data and test data is given in Fig. 5. The training and test accuracy is shown in Fig. 5a. PeachNet is trained for 500 iterations. The accuracy reaches 94%. The training and test loss for each iteration is shown in Fig. 5b. The loss decreases when more training is performed, and classification accuracy in different classes increases. The heatmap showing the precision, recall, and f1-score for each class in PeachNet is shown in Fig. 6. It can be observed that the identification of Peach Scab, Peach Leaf Curl, and Healthy Peach classes has low accuracy. The rest of the classes perform very well and achieved an accuracy of over 90%. The confusion matrix for classification by PeachNet is shown in Fig. 7a. The accuracy is high when the boxes on the diagonal have high values. It is observed that the diagonals of the matrix have higher values compared to the other regions. Here, 93% of the predicted labels are correct. The miss-class ratio is 0.0681. The ROC curve for PeachNet is shown in Fig. 7b. It indicates that the lines for different classes are on the left diagonal of the ROC, demonstrating reasonably good accuracy by PeachNet. The top-1 and top-5 errors with training time for CNN architectures is given in Tab. 4. The lowest training time is for LeNet, but the error rate is very high. Similarly, ResNet and RetinaNet have the lowest error rates but the highest training times. In our experiments, we used VGG-19 because it has reasonably good training time and an acceptable error rate. Conclusion and Future Work Food security is a crucial mechanism to protect food and crops from pests and diseases. We often lose food due to diseases and pests and, as a result, we are sometimes not able to meet the food demands of the market. Machine-learning techniques have been developed to protect the food supply from these threats and to increase the quality and quantity of food products. In this research, we experimented with a novel technique to detect diseases in peach fruit and trees. This method will help us to increase the quality and quantity of peaches grown around the world. The model was evaluated using several techniques, and the proposed technique has demonstrated an accuracy of 94%. The proposed model is the first step in automatic harvesting, and it can be extended in the future to recommend treatments once the class and location of disease are known.	2021-05-11T00:07:20.639Z	2021-01-01T00:00:00.000	{ "year": 2021, "sha1": "5a392bdc85c6059cac7b2e1849d11f75cb21ffa9", "oa_license": "CCBY", "oa_url": "https://doi.org/10.32604/cmc.2021.014950", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "f84fee2e4f083b1eabe97d7a3b51d32ec8b18ed3", "s2fieldsofstudy": [ "Agricultural And Food Sciences" ], "extfieldsofstudy": [] }
56433180	pes2o/s2orc	v3-fos-license	On Translation Strategies of the Epic Texts The paper deals with translation strategies of the epic texts in the case of the Sakha epic “Nurgun Botur the Swift” translation into English. Due to the variety of research works on the Sakha-Russian translation issues this paper focuses on the review of olonkho translations into other languages of the world. The framework theory includes the review of studies on translation strategies in Russia and abroad. As global translation strategies Lawrence Venuti’s ‘domestication’ and ‘ foreignisation’ are considered. The idea of “a golden mean strategy” is introduced. The paper views lexical translation challenges. Local strategies to overcome lexical challenges in the epic translation, as transliteration/ transcription, calque, partial calque, and explication occur to be the most applicable in the translation of the epic texts. Practical significance lies in the possibility of using the materials of this study by translators-practitioners in the translation of epic texts, in the courses on “Theory and practice of translation”, “Literary translation” as well as in assessing the quality of translations. Olonkho in the World's Languages Olonkho, a unique Sakha (Yakut) heroic epic, was proclaimed in 2005 a UNESCO Designated Masterpiece of the Oral and Intangible Heritage of Humanity.It has strongly marked specific genre features and sharply contrasts with other folklore genres in its ideological and thematic content, traditional range of images, stability of composition structure, originality of existence, olonkho performers' skills, and the traditional style and figures of speech (Emelyanov, 1980: 6). Olonkho is considered to be a specific field of culture, which reflects the identity, the culture and the mentality of the Sakha people.Therefore, it is rich in culture-specific vocabulary, epithets, word pairs, picturesque words and other peculiarities that complicate the process of translation.Professional responsibility of the translator is to overcome translation challenges applying different strategies. As noted by many olonkho researchers, the translation of the epic texts into the language of a different language group is extremely important and at the same time it is a difficult task.For instance, the Sakha language is of Altaic origin while the English language belongs to the Indo-European language family.(Kazakova, 2006: 317).Hence, the issue of epic texts translation strategy appears to be of great importance. Studies on Translation Strategies in Russia Translation strategy is a term that different researchers understand differently.On the one hand, it is a cognitive process that involves translation procedures that occur in the mind of the translator, on the other hand, a strategy implicates various translation techniques, methods.The common factor is that strategy is used by translators to find solutions to translation problems and implementation of adequate translation in general.The translator should evaluate the unity of his style without comparison with the original text, i.e. Strategies Abroad: Domestication and Foreignization In 1995, L. Venuti published a monograph "The Translator's Invisibility", where he first used the terms "domestication" and "foreignization", denoting the two main strategies by which the translator is guided in solving linguistic and cultural issues. Considering the translation strategies, L. Venuti introduces the concept of "invisibility" of the translator.L. Venuti defines the main characteristics of the translated text as ease of perception, i.e. the text is easy to read, and transparency, i.e. the text reflects the personality and intentions of the author, as a result, it seems that we read the original, not the translation. In this case, as L. Venuti emphasizes, the more easily the translation is read, the more invisible the translator is, and the more visible the author of the original is.Thus, the main characteristics of the translation strategy of domestication, according to L. Venuti, are fluency (ease of perception), transparency (transparency) and, as a consequence, the "invisibility" of the translator. As for the translations made in accordance with the translation strategy of foreignization, in this case certain "opacity" of the text arises, and the so-called "dark places" appear in it, the translator becomes "visible".This text is read as a translation (Venuti, 1995). According to Venuti, domestication is an ethnocentric approach, in which the original text is often reduced, the emphasis is on the cultural values of the target language, and the author is approaching the reader; foreignization is an approach in which the emphasis is on preserving foreign linguistic and cultural values, while "the reader is approaching the author".In a broad sense, domestication implies a "transparent", easy-to-understand style, due to which the foreign text appears less strange to the reader.In the case of foreignization, the text of the translation violates some generally accepted norms of the translation language, preserving the features of the original text. Venuti believes that domestication is aimed at compelling the target text to sound familiar to dominant culture and to express the concepts that are habitual for it.The strategy of domestication presupposes an ethnocentric approach, in which the source text is reduced and changed to suit the linguistic culture of the translation, which reflects the Eurocentric and, specifically, the Anglo-centric trend of the era of imperialism. The opposite approach, in its turn, leads to the need to eliminate the smoothing of the distinctive features of the foreign text. Thus, domestication strategy makes the translator "invisible", merely omitting notions and phenomena that require additional explanation, or replacing them with similar cultural phenomena of the target language.It contrasts with foreignization strategy, in which the translator becomes "visible", explaining and commenting on the untranslatable features of the foreign language text and culture (Venuti, 1995). There we would point out in advance that in the case when a translator seeks a "golden mean"; his "visibility" can be felt differently in different parts of the text. Foreignization involves application of translation methods emphasizing the distinctive features of the culture of the source text that the culture of the target language lacks.As a result, the translated text does not resemble either the original text or any other text in the target language.With foreignization, the reader is approaching the culture of the original (Venuti, 1995). Venuti indicated that it is preferable for the translator to reduce ethnocentric compulsion when translating.The strategy of foreignization allows limiting the forced "domestication" of world cultural values by the English-speaking part of the world.Thus, foreignization is aimed at emphasizing the features of the original text, and not hiding the "presence of a translator". This will ensure protection from the ideological pressure of the culture of the dominant language of translation.According to Venuti, domestication and foreignization are heuristic concepts.Their poles in translation vary depending on the time and status of the countries of the original texts and the translation.However, the fact that accentuation or ignoring the cultural peculiarities of a particular foreign language is due to the dominance/subordination of its culture in the world at the specific momentremains unchanged (Venuti, 1995). Global and Local Strategies As it was mentioned above, free and literal translations distinguished by Russian translation researchers may correlate with Venuti's strategies of domestication and foreignization. L.K. Latyshev, along with these two types of translation, distinguishes a "proper translation", which in its turn can be referred to as "the golden mean strategy".He writes that the proper translation is to a certain extent a hybrid of literal and free translations.From the literal, it took an orientation toward the maximal semanticstructural closeness to the original text, but only where it is possible, and does not lead to a violation of the norms of the target language, does not create ambiguities and inadequate perception of the addressee (Latyshev, 2003: 17). Indeed, in the modern translation practice, foreignization and domestication are extremely rare; none of the translations represent one or the other method "in its pure form".The third translation strategy, the so-called the golden mean strategy, is a successful combination of the two translational directions.In fact, each time it is a combination of opposing translation strategies.After all, simultaneous use of opposing translation methods leads to what is called the "golden mean", and ensures the achievement of artistic accuracy. Translator's choice of one or another strategy (foreignization, domestication, golden mean strategy) is called a global strategy, whereas local strategies deal with translation challenges that arise directly in the translation process.Such strategies can be seen as translation transformations, methods, or means.Table 1 presents the distribution of the most popular and frequent local strategies to overcome the lexical problems of translation by global strategies. Most Common Lexical Challenges of the Epic Translation The goal of the epic translation is not only to introduce a new culture and language of the source text to a foreign reader, but also to make the epic a heritage of world culture, which in its turn will contribute to raising the prestige of the whole ethnicity.Therefore, it is necessary to make the text interesting and readable.In other words, we believe the golden mean strategy to be an ideal strategy for epic texts translation.But what are the actual facts? This paper studies some lexical aspects in the translation of the heroic epic "Nurgun Botur the Swift" by A.A. Nakhodkina (Oiunskii, (2); Өлүү Чөркөчөөх -Oljuu Chorkochokh [ol'dʒu: 'tʃo:kotʃok] (1) - Eluu-Cherkechekh [el' ju: 'tʃǝ:kǝtʃǝk] (2).Nevertheless, it should be taken into consideration that R.Iu.Skrybykin's translation was based on the experience of the (Skrybykin and Oiuunuskay, 1995: 11) Where the air was light and blue, The sunny midday land was there, The creamy, milky lake was there.Each step he took brought him an ilgeh blessing, His breath was hot, He had plentiful supply of food, He lived surrounded by abundance… (Nakhodkina and Oyunsky, 2014: 6) Yakut-Russian translation that was established in the late 19 th century. The Sakha epic is rich in comparisons, metaphors, hyperboles, and epithets.They give the narrative a special colour and expressiveness. Most common are the epithets to the names of heroes, as well as to the descriptions of deities and worlds.These epithets sometimes contain an entire characteristic. This fragment describes the place where the supreme deity of the Upper World Urung-Aar Toyon lives.First of all, we should pay attention to the fact that R.Iu.Skrybykin does not allocate the beginnings of lines across the whole text. However, in the original text each line is marked by capital letters. Omission is found frequently in R.Iu. Skrybykin's olonkho.For instance, in this fragment a culture-specific word ilgeh (transcription of the term by Nakhodkinaauthor) is omitted.This word means "divine bestowed by the gods".In the second variant the word is transliterated and given with explanation blessing.The epithets to the lake that is given in the first variants as serene that means "peaceful and calm" does not refer to the Sakha words үрүмэтийбит үүт ([ürümǝtijbit ü:t] -author). The lake here is described as "foamed and milky". The second variant creamy, milky lake is more appropriate.Word pairs are also of great interest. Υүт-аас ([ü:t a:s] -author) is litteraly translated as "milk and food".It is known that milk and milk products played a great for the Sakha people because their life depended directly on the cattle. The more cattle you had, the richer and wealthier They can be divided by localization in the text: 1) in the translation text, next to realia; 2) outside the translation text, with reference to the select glossary of nontranslated words or translation commentaries, which contain a detailed interpretation of the realia, additional information on this concept; 3) without a reference word -the translation of a culture-specific word without the use of exoticism (Vasilyeva, 2011: 110). The translation of A.A. Nakhodkina tends to preserve culture-specific words.Therefore, the first type of explication when the words are accompanied by their English equivalents appears many times.For instance, "seleh rope", "kekhe hooks" (long hook fixed in the Yakut balagan house to hang clothes, kitchen ware, etc.), "choroon goblets", "matachakh vessels" (wooden vessels for keeping dairy food); "bagakh post" (a sacred post to which a sacrificial animal is tethered, or hung with the skin of such animal). As a result we have attributive word combination where nucleus is expressed by the English Conclusions Translation demands a deep understanding and a solid knowledge of both languages and cultures.Olonkho represents the ideas, the language and the philosophy of the Sakha people. Undoubtedly, olonkho should be studied, printed, translated into the languages of the world.This is one of the ways of our epic to become the property of world culture, and this in turn will contribute to raising the prestige of our entire ethnicity. Translation work requires meticulousness. The translator must be competent, have a good knowledge of vocabulary of both the source and target languages and consult dictionaries, The first time when the olonkho spoke in a foreign language was in a line-by-line translation into German in the appendix to the work of O. Boetlingk "On the Language of the Yakuts" (1851).The first translation of the olonkho into English was the translation of the olonkho of A.Ia. Uvarovskii "Er Sogotokh" carried out by Douglas Lindsey, an Associate Professor of Traumatic Surgery Department of Arizona University Medical College, and published in the quarterly edition of the University of Arizona in 1971.In 1984, Milos Krno translated the olonkho "Nurgun Botur the Swift" into the Slovak language.The author of several articles on the ethnography of Yakutia Yankel Karro together with L.M. Sabaraikina, a NEFU Associate Professor, translated K.G.Orosin's olonkho "Eles Botur" into French; he also translated a fragment of the olonkho "Nurgun Botur the Swift" of P.A. Oyunky for children.The first song of this olonkho was also translated into English by A.A. Skryabina in 1993 and by R.Iu.Skrybykin in 1995.In 2002, A.A. Skryabina made a fulltext translation of P.V. Ogotoev's olonkho "Eles Bootur".This olonkho was translated into French by a NEFU Associate Professor V.I.Shaposhnikova in 2015.In 2010, olonkho of K.G.Orosin "Nurgun Botur the Swift" spoke in Turkish, this work was performed by Murat Erzos.In 2014, "Nurgun Bootur the Swift" was translated into the Kyrgyz language by Sharshenaly Abdylbaev.A group of translators under the supervision of Doctor of Philology, Associate Professor A.A. Nakhodkina translated one of the most famous and voluminous olonkho of the Yakut folklore by P.A. Oiunskii "Nurgun Botur the Swift" into English in 2013.This is the full-text translation of the Yakut heroic epic into English.The publication was edited by Paul Norbury, Director of Renaissanse Books Publishing, Alina Nakhodkina, Doctor of Philology, Associate Professor, head of the NEFU Translation Department, Geneviève Perreault from Canada, including Doctor of Philology Svetlana Johnston, a native from Yakutia, living in the UK.In 2016, V.K. Alekseeva translated the olonkho of D.A. Tomskaia -Chaika "Kyys Kylaabinai the Warrior"; in 2017 "Ogo Tulaaiakh" by V. Karataev and "Khaptaljin Baatyr" by S. Chernogradskii.It was edited by NEFU professor V.V. Illarionov.As T.A. Kazakova notes, folklore texts represent a special difficulty as an object of translation.They are accompanied by a duality of choice, when the translator has to decide whether to preserve the specifics of the original formula in translation or to look for an analogue . Alekseeva understands the order and essence of the translator's actions when translating a specific text.At the same time, she highlights the importance of differentiating the concepts of "translation actions" and "translation strategy".Translation actions, according to the researcher, refer to the whole set of possible actions for the implementation of translation, while translation strategy is the consciously chosen algorithm of these actions by the translator when translating one particular text (or group of texts) (Alekseeva, 2008: 329).I.S. Alekseeva describes in detail three stages of translation strategies: pre-translation analysis, proper translation and post-translational processing of the text.A pre-translation analysis includes the collection of external information on the text, the definition of the source and the recipient, its communicative task and genre.Proper translation is a "search": the translator goes "analytically", each time "commenting on his actions and explaining the reason".The last stage is an analysis of the results of the translation and includes technical work on the translation text. V .N. Komissarov gives several points that determine the choice strategy: the nature of the source text(Komissarov, 2002: 72); the degree of prestige of the foreign author in the host culture; genre of literary work(Komissarov, 2002: 70).V.N.Komissarov also emphasizes the importance of socio-cultural influence on the translation strategy, which is often reflected in the completeness of the translation of the original, forcing the translator to reduce or completely omit the unacceptable in the receiving culture(Komissarov, 2002: 74).L.G.Shereminskaia also applies the term "translation strategy".She argues that mastering the correct strategy of the translation process is an important condition for the effectiveness of translation activities.It is based on a number of fundamental principles that are used at the conscious or unconscious level by each professional translator depending on the specific conditions of the translation action.Firstly, translation strategy, according to Shereminskaia, is an understanding of the subject, which is a necessary condition for achieving a good result.Secondly, it is the ability to determine the semantic dominant, the most important part of the content of the translated statement.The basic principles of the translation strategy are supplemented by the justification for the validity of a number of techniques (Shereminskaia, 2008: 62-64).T.A. Kazakova draws attention to the fact that communicative success to a great extend depends on how correctly the translator chooses the method of translation, applies the appropriate strategy and determines the units of translation (Kazakova, 2001: 11).Although the researcher does not give the definition of the term, she indicates that in the semantic translation two strategies naturally interact: a strategy oriented towards the method of expression accepted in the target language, and a strategy focused on preserving the features of the original form of expression (Kazakova, 2001: 14).In Russian translation studies, A. D. Schweitzer was one of the firsts to use the term "translation strategy".He points out that translation, in general, consists of a series of choices.At the first stage, the translator faces a choice of a translation strategy, which includes, first and foremost, the decision of translator to give preference to either a textually accurate (literal) translation or a translation greatly deviating from the formal structure of the source text, close to free.Afterwards, the translator decides on the aspects of the original, which must be reflected in the TT primarily.Thereby he creates a certain hierarchy of values, organizes prioritizes (Schweitzer, 1988: 65).A.D. Schweitzer's and T.A. Kazakova's dualism is close to two strategies of American translation theorists L. Venuti. you were.That was the reason why the translator of the second variant expanded the word pair үүт-аас and used the means of addition: He had plentiful supply of food, He lived surrounded by abundance.When translating from the Sakha language into English, the translator encounters the fact that the English-speaking tradition lacks many concepts associated with the mythology and religion of the Sakha people.Therefore, an explication is an obligatory component of the epic translation.This allows conveying the meaning of everyday realias, the realias of the world and nature, mythological places, and representatives of mythological communities.A.A. Vasilyeva points out that in the translation of the epic texts different types of explication are commonly used. equivalent and the transliterated culture-specific element as an attribute.Translation of epic texts is impossible without translation comments.It is known that a translation comment is a strategy of foreignization, since it makes the translator visible to the reader.It is a way to compensate for the loss in translation and provides additional information that cannot be integrated into the general translation text.It is used for the culturebearer to understand and feel the peculiarities of the other.A translation comment, usually given in the form of footnotes or comments, provides additional information that cannot be integrated into the general translation text.This second type of explication is also used by A.A. Nakhodkina, whereas R.Iu.Skrybykin's version does not contain footnotes; some explications of unknown to the English-speaking reader phenomena are given after the text.It is worth emphasizing that sometimes commentaries can be presented with some photos that illustrate in detail the most archaic realities.The third type of explication, the translation of a culture-specific word without the use of exoticism, is of rare occurrence in A.A. Nakhodkina's translation.For example, the word арбађас denoting 'a coat with shabby, faded fur' is translated as worn-out, ragged fur coat.It does not have an equivalent in the English language, so the translator opts to using the explication, thus achieving a better understanding.To create meaningful units in the translated text and to preserve the elements of the form or function of the original unit, the method of calque is used.For instance, the names of the three worlds of olonkho are translated by the calque: Аллараа дойду (Нижний мир) -the Under World, Yөһээ дойду (Верхний мир) -the Upper World, Орто дойду (Средний мир) -the Middle World.Partial calque of words and word combinations, consisting partly of the elements of the source language, partly from the elements of the language of the receiver, also commonly occurs.The combination of transliteration and calque is used in the translation of names of mythological heroes, containing semantic components, reflecting some of their real qualities.For instance, Улуутуйар Улуу Суорун Тойон -Great Uluutuyar Uluu Sorun Toyon; Күн тойон -the Sun -Kun Toyon; Ый хотун -the Moon -Yi Khotun; Чолбон тойон -The Venus -Cholbon Toyon; Υргэл тойон -the Pleiades -Urgel Toyon. thesauruses and other reference books to find the appropriate terms.The golden mean strategy can be considered ideal for translation epic texts.To be critical, the strategy should tend more towards foreignization than domestication.In the case of domestication, the epic will lose all its colorfulness and expressiveness.After all, the translation of the olonkho is the contribution of the Sakha people to the treasury of the world culture.Olonkho reflects a rich fantasy, a great idea of infinity, the triumph of life over death, and the translator's duty to bring all this to the reader as preserved as possible.Still, each translation job is unique and the choice of domestication, foreignization or golden mean as translation strategy depends on a concrete case and on the translation goal.As it was mentioned previously, I.S. Alekseeva defines three stages of translation strategies: pre-translation analysis, proper translation and post-translational processing of the text.Regarding translation stages/strategies of the epic texts we can distinguish the following: firstly, the pre-translation analysis, which implies the search for extralinguistic information, studying linguistic, cultural and ethnic peculiarities, the identification of cultural and linguistic realities, etc.; secondly, the definition of main difficulties in translation (phonetic, phonosemantic, semiotic, syntactic, etc.); thirdly, the proper translation with maximum preservation of the identity of the text, with an accurate transfer of the semantic content of the olonkho in compliance with the norms of translation; fourthly, technical work on the translation product including proofreading, check for collocations and grammar issues by a native speaker.Life does not stand still, the translation is developing.It is delightful that every year thetranslations of olonkho into the languages of the world become more.As we have seen, the quality of translation is also growing.	2018-12-17T23:23:27.372Z	2018-01-01T00:00:00.000	{ "year": 2018, "sha1": "8216f8ff931049fda7379ee91a6934359f352085", "oa_license": "CCBYNC", "oa_url": "http://elib.sfu-kras.ru/bitstream/2311/71364/1/Dyachkovskaya.pdf", "oa_status": "HYBRID", "pdf_src": "Anansi", "pdf_hash": "8216f8ff931049fda7379ee91a6934359f352085", "s2fieldsofstudy": [ "Linguistics" ], "extfieldsofstudy": [] }
238582906	pes2o/s2orc	v3-fos-license	Depth Optimized Ansatz Circuit in QAOA for Max-Cut While a Quantum Approximate Optimization Algorithm (QAOA) is intended to provide a quantum advantage in finding approximate solutions to combinatorial optimization problems, noise in the system is a hurdle in exploiting its full potential. Several error mitigation techniques have been studied to lessen the effect of noise on this algorithm. Recently, Majumdar et al. proposed a Depth First Search (DFS) based method to reduce $n-1$ CNOT gates in the ansatz design of QAOA for finding Max-Cut in a graph G = (V, E), \|V\| = n. However, this method tends to increase the depth of the circuit, making it more prone to relaxation error. The depth of the circuit is proportional to the height of the DFS tree, which can be $n-1$ in the worst case. In this paper, we propose an $O(\Delta \cdot n^2)$ greedy heuristic algorithm, where $\Delta$ is the maximum degree of the graph, that finds a spanning tree of lower height, thus reducing the overall depth of the circuit while still retaining the $n-1$ reduction in the number of CNOT gates needed in the ansatz. We numerically show that this algorithm achieves nearly 10 times increase in the probability of success for each iteration of QAOA for Max-Cut. We further show that although the average depth of the circuit produced by this heuristic algorithm still grows linearly with n, our algorithm reduces the slope of the linear increase from 1 to 0.11. Introduction Quantum Approximate Optimization Algorithm (QAOA) [1] is a hybrid quantum-classical algorithm [2], studied primarily for finding an approximate solution to combinatorial optimization problems. A QAOA is characterized by a Problem Hamiltonian H P that encodes the combinatorial optimization problem, and a Mixer Hamiltonian H M that anti-commutes with H P and whose ground state is easy to prepare. Two parameterized unitaries U (H P , γ) = exp(−iγH P ) and U (H M , β) = exp(−iβH M ) are applied sequentially for p ≥ 1 times on the initial state \|ψ 0 . Here γ = {γ 1 , γ 2 , . . . , γ p } and β = {β 1 , β 2 , . . . , β p }, γ i , β i ∈ R ∀ i, are the parameters. A single epoch of a level-p QAOA is represented as in Eq. (1). The parameters are initialized randomly, and after an epoch of iterations, that provides the expectation value ψ(γ, β)\|H P \|ψ(γ, β) , the parameters are updated by a classical optimizer. The next epoch uses this new set of parameters and is expected to provide an expectation value that is closer to the optimum solution to the problem. Farhi et al. first proposed QAOA [1], and studied it in the context of finding a maximum cut in a graph, known as the Max-Cut problem. For 3-regular graphs, they showed that a p = 1 QAOA achieves an approximation ratio better than random guessing, but lower than the best known classical algorithm [3]. They also argued that the expectation value of the cut produced by the algorithm is a non-decreasing function of p. Therefore, QAOA is expected to be a potential candidate for quantum advantage using near-term devices. Many researchers, since then, have studied QAOA in the context of the Max-Cut problem [4][5][6][7][8][9]. A recent experiment from Google [10] showed that noise overwhelmed QAOA for Max-Cut in current quantum devices, and the expectation value produced by the algorithm decreases beyond p = 3. Error mitigation processes have been studied in the literature that lowers the effect of noise on QAOA, or in general on hybrid quantum-classical algorithms [11][12][13]. Apart from error mitigation techniques, variation in the Mixer Hamiltonian [7], or the Cost Function [6,14] have been proposed that either lowers the noise in the circuit, or achieves faster convergence, thus reducing the depth, and hence the effect of decoherence, of the circuit. In [9], Majumdar et al. proposed a Depth First Search (DFS) based method that can eliminate n − 1 CNOT gates in the circuit of QAOA Max-Cut for any graph G = (V, E), where \|V \| = n. Since CNOT gates are one of the primary sources of error in modern quantum devices [15], this procedure significantly reduces the noise in the circuit. As this method imposes an ordering of the edges (discussed in detail in Sec. 2), there is an increase in the depth of the circuit. The authors, however, proved that this increase in depth is overshadowed by the reduction in CNOT gates, and the overall circuit has a lower probability of error. Nevertheless, a graph has multiple DFS trees, and the depth of the circuit varies with the height of a DFS tree for the given graph. A circuit with lower depth is naturally preferable with effect of decoherence being lower. Contributions of this article It is not a trivial task to find a DFS tree for a given graph that is guaranteed to provide a low depth ansatz circuit. Further, a DFS tree with lower height does not necessarily result in a lower depth circuit (see Sec. 3). In this paper, we propose a greedy heuristic algorithm to lower the depth of the circuit while still eliminating n − 1 CNOT gates. The run-time of our proposed algorithm is O(∆ · n 2 ), where ∆ is the maximum degree of the graph, and it reduces the depth of the circuit by 84.8% for graphs with number of vertices n = 100. We show our results on Erdos-Renyi Graphs with the probability of edge varying from 0.4 -0.8, and complete graphs. For graphs with 4 ≤ n ≤ 12, we simulate our proposed heuristic with the ibmq manhattan noise model, and show more than 10 times increase in the success probability of each iteration of QAOA for Max-Cut. The maximum height of the DFS tree (and hence the depth of the circuit) can be n − 1, i.e., it increases linearly with the number of vertices with a slope of 1. We show that although the increase in depth of the circuit with the number of vertices from our method is still linear, the slope is reduced to 0.11. In the rest of the paper, Sec. 2 gives a brief review of QAOA for Max-Cut and the DFS based optimization proposed in [9]. In Sec. 3 and Sec. 4 we respectively provide the conditions that lead to a low depth circuit, and a corresponding greedy heuristic algorithm to achieve a low depth. Sec. 5 presents the simulation results of this proposed algorithm, and the concluding remarks appear in Sec. 6. 2 QAOA for Max-Cut and DFS based ansatz optimization QAOA for Max-Cut The traditional QAOA ansatz (the parameterized circuit) is composed of the two parameterized unitaries U (H P , γ) and U (H M , β). Given a graph G = (V, E), with \|V \| = n, the circuit is initialized in the equal superposition of n qubits. This initialization step has a depth of 1 (simultaneous operations of Hadamard gate on all qubits). Similarly, the circuit realization of U (H M , β) is a simultaneous operation of R X (β) on all the qubits. The operator U (H P , γ) depends on the Problem Hamiltonian. For the Max-Cut problem, . The operator U (H P , γ) can, therefore, be represented as in Eq. (2). The circuit realization of the operator U (H (j,k) P ) corresponding to an edge (j, k) is shown in Fig. 1. In accordance to the nomenclature used in [9], we call this circuit a step. More precisely, multiple simultaneous such operators can be executed in each step corresponding to disjoint edges. For example, in For the rest of the paper, the term QAOA will refer to the QAOA for the Max-Cut problem where the graph G = (V, E) is connected, undirected, and unweighted, and \|V \| = n and \|E\| = m. Every analysis, heuristic and algorithm in this paper will be applicable directly or with minimal changes to weighted graphs or graphs with multiple components. Furthermore, the term an edge is operated on, or its equivalent terms, would imply the operation of the circuit of Fig. 1 for that corresponding edge. For the sake of completeness, in Fig. 2 we show the circuit for a p = 1 QAOA corresponding to a 2-regular graph with four vertices. DFS based optimization of the ansatz circuit We now briefly discuss the DFS based optimization method that was proposed in [9]. Given a graph G = (V, E), this method finds a DFS tree T starting from a randomly chosen root vertex r. The DFS tree is an acyclic subgraph of G containing all the n vertices and n − 1 edges. The QAOA circuit corresponding to U (H P , γ) can then be partitioned into two portions -the one with edges in T , and the other edges. The operators corresponding to the other edges, which are not included in the DFS tree, can be executed in any order once all the operators corresponding to the edges in T are operated on. For each edge (u, v) ∈ T , the operator U (H (u,v) P ) is operated on maintaining the following conditions: i) Starting from the root vertex r, if an edge e 1 appears earlier than another edge e 2 in the DFS tree T , then the operator U (H e1 P ) must precede U (H e2 P ) in the corresponding circuit. ii) If an edge e = (u, v) is included in T and the vertex u is incident on an edge already in T , then u and v must act as the control and target of the CNOT gate respectively in the operator U (H e P ). Majumdar et al. [9] proved that if these two conditions are satisfied, then the first CNOT gate for the operator in Fig. 1, associated with each edge in T , can be removed while still retaining functional equivalence for p = 1 of the QAOA circuit. This optimization method does not hold for p > 1. Nevertheless, for any level−p QAOA, the first level can be thus optimized. Therefore, the DFS based method can reduce the CNOT count of the overall QAOA circuit by n − 1. This method mandates a sequential ordering of the tree edges, i.e., an edge in T can be operated only after operators for all of its ancestors have been applied. The maximum height of a DFS tree with n vertices can be n − 1, thus leading to a significant increase in the depth of the circuit. Fig. 3 shows a 2-regular cycle with 6 vertices. In the traditional QAOA ansatz, the operator U (H P , γ) can be executed in 2 steps only. As indicated with the two colors in the graph of Fig. 3 (a), the edges assigned the same color can be operated on in the same step. The DFS based method on the other hand (the graph of Fig. 3 (b)) requires 6 steps. It is easy to verify that for a 2-regular cycle, the traditional QAOA ansatz always requires 2 (for even n) or 3 (for odd n) steps only, whereas the DFS based optimized circuit requires O(n) steps. For large graphs, it is possible that the depth of the DFS based optimized circuit leads to an execution time greater than the coherence time of the hardware. Even when such a scenario does not occur, increase in depth makes the circuit more susceptible to decoherence. Therefore, although the authors in [9] showed that reduction in the numner of CNOT gates in the ansatz overshadows the increase in depth with respect to probability of error, a pertinent question is whether a DFS tree that reduces the depth of the circuit as well can be found efficiently. (a) In the traditional QAOA ansatz edges corresponding to the same color can be operated on in the same step; the number of steps needed for this example is 2. (b) In DFS tree based ansatz optimization -the number of CNOT gates can be reduced by 5, but the number of steps required is 6. Note that this approach to optimizing the number of CNOT gates holds for any rooted spanning tree where the two conditions mentioned above, on the ordering of the tree edges, are met. DFS is merely a method to generate a rooted spanning tree of a graph. Henceforth, instead of specifically finding the DFS tree, we focus on finding a rooted spanning tree for a graph G which serves the above purpose. Formulation of the Ansatz Optimization Problem Here we focus only on optimizing for the operator U (H P ) associated with each edge because each of the circuits for initialization and the Mixer Hamiltonian having a depth of one, is the same as for the traditional QAOA circuit [1], or for the optimized circuit in [9]. Henceforth, when we mention a circuit, or its depth, we refer to the circuit corresponding to the operator U (H P ) only. The maximum height of a DFS tree with n vertices is n − 1, and so is the depth of the corresponding circuit. It may be claimed that a spanning tree with lower height can lead to a circuit with lower depth. In such a case, a Breadth First Search (BFS) tree may provide a spanning tree with minimum height. However, the two trees shown in Fig. 4 (a) and (b) have different heights leading to the same depth of the circuit. In both the figures, the values associated with the edges depict the level at which the operator corresponding to that particular edge can be operated on so that the optimization (i.e. reducing the number of CNOT gates) holds. The circuit corresponding to both of the trees, shown in Fig. 5, is the same. These two trees and their corresponding circuit readily show that simply reducing the height of the tree is not sufficient to obtain a circuit of lower depth. Therefore, finding a BFS tree instead of a DFS tree does not guarantee a circuit with lower depth. Figure 5: The quantum circuit of U (H P , γ) corresponding to both the trees in Fig. 4 The reason why the tree on the right hand side with height 2 in Fig. 4 cannot lower the depth of the circuit is because the CNOT gates corresponding to two adjacent edges cannot operate at the same step. Therefore, simply reducing the height of the spanning tree is not sufficient to reduce the depth of the circuit. Furthermore, in Fig. 6, we show two trees with the same height, but having circuits of different depth, as evident from the level number attached with the edges. We now define a few terms to easily clarify the requirements for a rooted spanning tree that will lead to a circuit with lower depth. 1. Branching Factor: The branching factor of a vertex v is defined as the number of vertices that have been discovered in the rooted spanning tree from v. In other words, the branching factor of a vertex v is one less than the degree of that vertex in the spanning tree except for the root vertex, whose branching factor is equal to its degree. For example, in the tree of Fig. 4 (b), starting from the root labelled 1, the branching factor of the root is 1, that of vertex 2 is 2, and that of the leaf vertices are 0. We see an example of delayed start in the tree of Fig. 4 (b). Although both the leaf vertices in that tree are in the same level, they cannot be operated on simultaneously. Therefore, they must be operated on two disjoint levels. Indeed delayed start is the reason that the depth of the circuit does not reduce directly with the height of the tree. It is obvious that a tree with a higher branching factor can experience more delayed start than a tree with a lower branching factor. On the other hand, the height of the tree increases with decreasing branching factor. Therefore, we need a spanning tree starting from a root vertex r that has as few delayed starts as possible. This rooted spanning tree is neither a DFS tree that has low branching factor, nor a BFS tree that has low height. It is, rather, a tree that has a trade-off between the branching factor (and hence delayed start) and the height. Conjecture: The problem is NP-Complete We have seen that simply finding a rooted spanning tree with minimum height is not sufficient to reduce the depth of the corresponding circuit. What we need instead is to have edges that can be executed in parallel. This is similar to the Edge Coloring problem [16]. Edges having the same color are disjoint and can be executed in parallel. This method was exploited in [9] as well. However, the problem here is more constrained than the Edge Coloring problem. In a tree, the same color can be used for edges in alternate levels, i.e., it is possible to have edges of same color in level 1, level 3 etc., and in level 2, level 4 etc. For example, in Fig. 4 (a), normal Edge Coloring assigns the same color to the edges in levels 1 and 3 since they are disjoint. However, we have already discussed that operators corresponding to these two edges cannot execute in parallel. Therefore, we have an added constraint that an edge cannot be given the color of any of its ancestors. We formally define the problem as follows: Problem 1. Given a graph G = (V, E), starting from a root vertex r find a spanning tree T of G whose edges can be colored with the minimum number of colors satisfying the conditions that i) No two edges incident on a common vertex have same color. ii) No edge has same color as that of any of its ancestors. Optimal Edge Coloring problem is itself an NP-Complete problem, and Problem 1 has additional constraints. In other words, we want that the degrees of the vertices in the spanning tree are not very large in order to avoid delayed start. Edges which are suffering from delayed start due to the rooted spanning tree ordering cannot have the same colors, and will thus increase the required number of colors. In [17], the authors showed that finding a degree constrained spanning tree, i.e., a spanning tree where the degree of any vertex is upper bounded by a predefined value, in NP-Complete. Hence, we conjecture that Problem 1 is NP-Complete, and propose a greedy polynomial time algorithm to find a better solution instead of the vanilla flavour depth-first search based method. Proposed Cost Function and Algorithm We propose a cost function that respects the following observations: i) If the branching factors of the vertices are very high, then the corresponding circuit will suffer from delayed start, leading to an increase in the depth. On the other hand, if the branching factor of the vertices are very low, then the height of the tree, and hence of the depth of the circuit, will increase. ii) Between two vertices u and v, it is better to branch the one at a lower level of the tree so that the edges in that branch may still have some opportunity to be executed in parallel with other edges at a higher level even after delayed start. An example of this is shown in Fig. 6 where both the trees have the same height, but the tree of Fig. 6 (b) will lead to a circuit with lower depth since the branching is closer to the root. iii) For graphs with fewer vertices, the branching factor should be low in order to avoid increase in depth due to delayed start. However, as the number of vertices increases, a higher branching factor must be allowed to lower the height of the tree. Respecting all the three criteria stated above, we propose a cost function C v to be associated with every vertex v. Let n be the number of vertices in the graph, l v and v bf be the level and the current branching factor of the vertex v respectively, and B be the maximum branching factor decided for any vertex in the spanning tree, then When growing the spanning tree from a root vertex, the edge (v, w) for which the cost function C v is maximum, is added to the tree. Note here that for a new edge (v, w), the cost function does not depend on the vertex w, but rather on the vertex v from which this edge is discovered (the algorithm is provided later on). The term (n − l v ) is always positive. On the other hand, Our proposed algorithm avoids branching at a vertex for which v bf ≥ B. In fact, when v bf = B, the cost function has a contribution of 0. Therefore, in our result section, we take B = f + 1 if we want a maximum branching factor of f in the spanning tree. Furthermore, the term (n − l v ) is higher for the vertices with lower l v . Thus, if for two vertices u = v, v bf = u bf < B, the algorithm chooses to branch that vertex which has a lower level. This ensures that delayed start is closer to the root, so that those branches still have some opportunity for parallel execution with some higher level branches. Furthermore, if v bf > B, the product with (n − l v ) leads to significantly low values for low l v . This discourages branching more than B in lower levels of the tree strongly to prevent excessive delayed start (like in a BFS tree). In other words, the spanning tree generated by this heuristic cost function is neither a BFS nor a DFS one, but rather an intermediate one. Algorithm 1 presents how to generate a spanning tree that has a trade-off between the height of the tree and the branching factor. 19: Algorithm 1 Cost Function Based Rooted Spanning Tree Generation x bf = x bf + 1. 20: Remove all edges of the form ( * , q) from edges to add. 22: if q / ∈ V isited then 23: edges to add = edges to add ∪ {edge}. Proof. Let r be the randomly chosen root vertex of the spanning tree. Therefore, the choice of root does not require any computational time. Since ∆ is the maximum degree of the graph, r can have at most ∆ neighbours. Finding the maximum cost function among these neighbours require O(∆) time. Subsequent vertices in the spanning tree can have at most ∆ − 1 neighbours since one of its neighbour must be its parent in the spanning tree. Therefore, the total time requirement in all the steps is W ≤ ∆ (to create the spanning tree upto two vertices) ≤ ∆ + (∆ − 1) (to create the spanning tree upto three vertices) ≤ 3∆ − 2 (to create the spanning tree upto four vertices) . . . Therefore, For sparse graphs, ∆ = O(1) and for dense graphs ∆ = O(n). Therefore, the time complexity of the proposed Algorithm 1 varies between O(n 2 ) to O(n 3 ) depending on the sparsity of the given graph. An Illustration of Algorithm 1 We illustrate the DFS method [9] and our proposed method in action on an example graph given in Fig. 7. First, in Fig. 8 we show the traditional p = 1 QAOA circuit for this graph. Then, in Fig. 9 we give two spanning trees of the graph. The spanning tree in Fig. 9 (a) is generated using the DFS method [9], whereas the one in Fig. 9 (b) is generated using Algorithm 1 with B = 3. In Fig. 10 (a) and (b) we show the optimized circuits for the p = 1 QAOA of the graph in Fig. 7, where the optimized circuits are generated using the DFS method [9] and Algorithm 1 respectively. The values of γ and β are randomly selected. Fig. 7 The depth of the circuits in Fig. 8, Fig. 10 (a) and Fig. 10 (b) are 11, 14 and 12 respectively as obtained using the .depth() function of Qiskit [18]. The number of CNOT gates in both the optimized (a) A spanning tree tree of the graph in Fig. 7 (b) Another spanning tree tree of the graph in Fig. 7 Figure 9: Two spanning trees of the graph in Fig. 7. The left graph is generated using the DFS method [9], and the right one is generated using Algorithm 1 with B = 3. (a) Optimized circuit using DFS method (b) Optimized circuit using proposed Algorithm 1 Figure 10: Optimized p = 1 QAOA circuit corresponding to U (H P , γ) for the two spanning trees in Fig. 9 respectively circuits in Fig. 10 are 5 less than that in Fig. 8. We note that the depth of both the optimized circuits are greater than that of the traditional QAOA. However, the optimized circuit in Fig. 10 (b) can be considered to be superior since it requires 5 CNOT gates fewer than that in Fig. 8, as well as increases the depth by 1 only. In Sec. 5, we show that our proposed Algorithm 1 can significantly arrest the increase in depth, and in some cases can lead to a lower depth than its traditional circuit. Reduction in the depth of the circuit The entire circuit of U (H P , γ) can be divided into two disjoint parts -one corresponding to the edges in the spanning tree, followed by the other edges in the input graph. Our algorithm can reduce the depth of the circuit corresponding to the spanning tree only. The circuit for the unoptimized edges remains the same as in [9]. Furthermore, the initialization, and the Mixer Hamiltonian is the same for the traditional QAOA circuit [1] or the circuit proposed in [9], and our proposed optimized circuit. Therefore, here we compare the depth of the circuit corresponding to the spanning tree only. When executing a circuit in a hardware, the graph has to be mapped to the underlying hardware connectivity graph. This process is called transpilation. All the results in this section are generated after transpiling the original circuit in the ibmq manhattan connectivity graph using the transpilation procedure of qiskit [18] with optimization level = 3. In the worst case, the height of the DFS tree, and hence the depth of the corresponding circuit, can be as large as n − 1. In Fig. 11 (a)-(d) we show the reduction in depth of of the circuit of the spanning tree by our proposed algorithm compared to the worst case depth of the circuit corresponding to the maximum height of the DFS tree. Fig. 11 (a)-(d) show the reduction in depth for Erdos-Renyi graphs with p edge , the probability of an edge, varying from 0.4 to 0.8, and complete graphs. For each type of graph, we vary n, the number of vertices from 20 to 100, and the value of the depth corresponding to each n is an average over 80 graph instances. The graph instances are same for all the values of B. The graphs in Fig. 11 and later in Fig. 12 are averaged over all the possible n spanning trees generated by selecting each of the n vertices once as the root. For all the types of graphs considered, we observe higher reduction in the depth for a higher value of B as the number of vertices increases. This is at par with our reasoning earlier, that for larger graphs, it is better to allow higher values of B. We have shown results for B = 3, 6 and 10 only. For B = 10 and n = 100, the reduction in the depth is 84.8%. It is evident from the graphs that the depth decreases with increasing B as the number of vertices increases. So for larger graphs, one should opt for even higher values of B. We observe from the graphs in Fig 11 that the increase in the depth with n for various values of B is still linear. In the worst case, where the depth is n − 1 for a graph with n vertices, the slope is 1. In Table 1 we show the slopes of the curves for B = 3, 6 and 10 for each of the graph family considered. From the values it is evident that the slope corresponding to the increase in depth is lowered by 1 10 as the value of B increases. Increase in the Probability of Success QAOA consists of executing the same circuit with the same parameters multiple times to obtain an expectation value of the cut. The performance of the algorithm is determined by this expectation value of the obtained cut. However, since our QAOA circuit, and the QAOA circuit in [9] are functionally equivalent to the traditional QAOA circuit, the performance remains unchanged. In this paper, we define success in a different way. For each iteration of the algorithm, let \|ψ denote the ideal state vector obtained via noiseless simulation. As real-world quantum devices are noisy, let \|ψ e denote the noisy outcome obtained via noisy simulation. We define the probability of success P success = \| ψ\|ψ e \| 2 . For a graph with n vertices, the optimization proposed in [9] reduced the number of CNOT gates by at most n − 1. Since a CNOT gate is one of the most acute sources of error, the method improved P success . However, this improvement has the overhead of increase in the depth, which exposes the circuit to more decoherence. In this paper, we have retained the n − 1 reduction in the number of CNOT gates needed for the operator U (H P ) in the ansatz and have also arrested the increase in depth to a bare minimum (refer Fig. 11 (a)-(d)). This leads to a further improvement in P success . The plots corresponding to P success is somewhat indecipherable because the decrease in P success with increasing n is significantly less in our method than the traditional QAOA or the optimization of [9]. Thus the changes are not easily observable in a plot for P success . Therefore, in Fig. 12 (a)-(d) we plot (1 − P success ) = 1 − \| ψ\|ψ e \| 2 for the four graph families using the ibmq manhattan noise model for the noisy simulation. We show that our proposed method decreases (1 − P success ) by more than 10 times as compared to the traditional QAOA or the optimization in [9] for Erdos-Renyi graphs with 0.4 ≤ p edge ≤ 0.8 and complete graph. We observe that in Fig. 12 (a)-(d), it is not evident that any betterment is achieved by increasing the value of B. However, this is simply because for graphs with low values of n, the maximum degrees are low as well. This is supported by similar results observed for low values of n in Fig. 11 (a)-(d) as well. However, we see in those plots that as n increases, increasing the value of B leads to a better result. So we expect to see better results for P success as well with increasing value of B as the number of vertices is increased further. Conclusion In [9] the authors showed that n − 1 CNOT gates can be omitted from the ansatz circuit of QAOA for Max-Cut for an n vertex graph if a DFS based ordering is followed while constructing the circuit. However, this led to an increase in the depth of the resulting circuit. In this paper, we have proposed a polynomial time heuristic algorithm that can find a rooted spanning tree such that the reduction in the number of CNOT gates is retained while significantly arresting the increase in depth. Our method is able to reduce the increase in depth by 1 10 as compared to the circuit in [9]. This, in its turn, leads to a significant increase in the success probability of the algorithm, since (i) the reduction in CNOT gates is retained, and (ii) the increase in depth is lowered thus making the circuit less susceptible to relaxation error. Our proposed heuristic for circuit synthesis is thus expected to provide a novel scheme for mitigating the effect of error in QAOA for Max-Cut, and can be used together with other error mitigation scheme. Our results show that higher branching factor is better to arrest the increase in depth as the number of vertices increases. A future prospect can be to study an approximate value of the branching factor for a particular n such that the increase in depth is minimum.	2021-10-12T01:34:09.940Z	2021-10-09T00:00:00.000	{ "year": 2021, "sha1": "f539e2db82dffc30b32a3b6640badfd0087145dd", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "f539e2db82dffc30b32a3b6640badfd0087145dd", "s2fieldsofstudy": [ "Computer Science", "Physics", "Engineering" ], "extfieldsofstudy": [ "Physics", "Computer Science" ] }
89946264	pes2o/s2orc	v3-fos-license	Characterization of partial resistance to black spot disease of Rosa sp. Black spot disease, caused by the fungus Diplocarpon rosae Wolf, is one of the most serious diseases of garden roses. Both complete (vertical) resistance conditioned by dominant Rdr genes and partial (horizontal) resistance conditioned by multiple genes have been described. The use of resistant rose cultivars would reduce the demand of agrochemical applications. The characterization of 16 genotypes for resistance to black spot using two laboratory assays, the detached leaf assay (DLA) and the whole plant inoculation (WPI) approach, indicated that these techniques were well correlated. Thus, either method could be used to assess the resistance of the plants to black spot. Fifteen diploid hybrid populations from 10 parents segregating for partial (horizontal) resistance to black spot derived from Rosa wichuraiana ‘Basye’s Thornless’ (RW) were assessed for black spot resistance by quantifying the percentage of the leaf area with symptoms (LAS) and lesion length (LL) measured by the diameter of the largest lesion per leaf in DLAs. The narrow-sense heritability of partial resistance to black spot as measured by LAS and LL data of DLA was estimated to be from 0.28 to 0.43 when calculated with a genetic variance analysis and from 0.74 to 0.86 when generated from offspring–midparent regression. This suggests that the development of rose cultivars with high levels of stable partial resistance to black spot is a feasible approach for the rose industry. Roses, which are distributed throughout the temperate regions of the Northern Hemisphere (Kr€ussmann, 1981), have been important ornamental plants for more than 5000 years. There are thousands of cultivars for the garden, floriculture, medicinal, fragrance, and culinary industries (Gudin, 2000; Marriott, 2003; Shepherd, 1954). The rose industry contributes a value of about $400 million from garden roses, which is the major crop in the $2.81 billion wholesale U.S. shrub market (AmericanHort, 2014). Compared with the rose market in the United States 35 years ago, the production of garden roses has decreased (Byrne et al., 2010) from 40 million roses down to 12 million field-grown and 15–18 million pot-grown rose bushes in 2012 (Hutton, 2012). This is thought to be because many rose cultivars have low tolerance to disease and abiotic stress (Waliczek et al., 2015). In response to a consumer demand for low-maintenance roses, many rose breeding programs are working toward developing cultivars resistant to the common rose diseases (Byrne, 2015; Debener and Byrne, 2014). Black spot, the most important disease affecting garden roses globally, is caused by the fungus D. rosaeWolf (Marssonina rosae anamorph) (Nauta and Spooner, 2000). The typical symptoms of this disease include dark rounded spots with a feathery edge on the adaxial side of the leaves while the abaxial epidermis remains unaffected. The disease can lead to the development of chlorosis around the lesion and eventually defoliation (Blechert and Debener, 2005; Gachomo et al., 2006; Horst and Cloyd, 2007). Eleven unique races of D. rosae have been identified among the isolates obtained from North America and Europe (Whitaker et al., 2010b). Two types of disease resistance to black spot have been characterized in roses. Vertical or complete resistance, which blocks sporulation and severely restricts the mycelial growth of the pathogen, is usually controlled bymajor dominant genes (Rdr or Rosa disease resistance genes) (von Malek and Debener, 1998; Whitaker et al., 2010a; Yokoya et al., 2000). In rose, the dominant resistance genes are pathogen race-specific, indicating a gene-for-gene interaction pattern (von Malek and Debener, 1998). Partial or horizontal resistance that appears to be non-race-specific has also been identified in roses (Shupert, 2005; Whitaker and Hokanson, 2009a, 2009b; Xue and Davidson, 1998). This resistance does not prevent infection of the pathogen, but rather delays disease development and results in reduced lesion size, reduced sporulation, and/or delayed infection after inoculation (Parlevliet, 1981; Whitaker and Hokanson, 2009a, 2009b; Xue and Davidson, 1998). Compared with complete resistance, partial resistance is generally more durable over the range of pathogenic races (Noack, 2003). The ideal disease-resistant genotype should have both highly effective and long-lasting resistance to a broad spectrum of pathogenic races (Blechert and Debener, 2005), which can be achieved by pyramiding dominant complete resistance genes, obtaining strong partial resistance or by combining both types of resistances. Black spot resistance is commonly evaluated in field trials in different geographic regions to expose the rose to a greater number of pathogenic races. These trials typically last 2–3 years to ensure sufficient disease pressure to properly assess the resistance of the plants (Carlson-Nilsson, 2002; Debener and Byrne, 2014; Noack, 2003; Shupert, 2005). The laboratory-based DLA is a tool for observing disease development efficiently under uniform and well-controlled environmental conditions and inoculum levels (Hattendorf et al., 2004; von Malek and Debener, 1998; Whitaker and Hokanson, 2009a, 2009b). Because single-conidial isolates are used in laboratory screening, the combination of compatible and incompatible interactions that are caused by various races in nature can be avoided (Blechert and Debener, 2005). However, as DLA only allows one cycle of disease development, the differences among genotypes might not be as accentuated as compared with a field trial in which multiple cycles of pathogen development are common (Xue and Davidson, 1998). Other factors such as the physical status of the host plant, degradation of the leaves, missing observations on leaf abscission/defoliation in DLA, and low or nonuniform inoculation levels in field assessment could all cause low correlation between these two methods of Received for publication 12 Sept. 2016. Accepted for publication 17 Nov. 2016. This work was partially funded by Monsanto Fellows in Plant Breeding program, the Robert E. Basye Endowment in Rose Genetics, and the USDA’s National Institute of Food and Agriculture (NIFA) Specialty Crop Research Initiative project, ‘‘RosBREED: Combining disease resistance with horticultural quality in new Rosaceous cultivars.’’ We thank the Hokanson lab at the University of Minnesota for supplying race 8 of Diplocarpon rosae used in this work. Corresponding author. E-mail: dbyrne@tamu.edu. HORTSCIENCE VOL. 52(1) JANUARY 2017 49 phenotyping (Johansson et al., 1992; Palmer et al., 1966; Zlesak et al., 2010). The objectives of this research were to 1) evaluate two methods (DLA and WPI) of artificial inoculation for black spot disease evaluation and characterize rose genotypes for black spot resistance and 2) characterize partial black spot disease resistance derived from RW in laboratory tests in diploid populations to estimate the components of genetic variances, and the heritability of this partial resistance. Materials and Methods Plant materials. Based on previous field ratings for black spot resistance, seven black spot–susceptible roses [‘Cal Poly’ (Cal), ‘Golden Gardens’, ‘Orange Honey’ (OH), ‘Red Fairy’ (RF), ‘Sweet Chariot’ (SC), ‘Vineyard Song’ (VS), and ‘Violette’], one with moderate resistance (‘Old Blush’), seven black spot–resistant breeding lines (91/100-5, DD, FF, J06-20-14-3, J06-28-46, J06-30-3-6, M4-4, and one rose species RW) were propagated from cuttings and grown in 1-gal pots containing a growth media of decomposed pine bark amended with Metro-Mix growing media (Sun Gro Horticulture Canada CM Ltd, Agawam, WA) under the greenhouse environment for 3 months before the experiments (Byrne et al., 2010; Zlesak et al., 2010). Nine individuals were randomly selected from each genotype for artificial inoculation for black spot resistance evaluation. The experiment was repeated three times. Fifteen diploid populations were generated in an incomplete diallel mating design by crossing five black spot–resistant lines [J06-20-14-3 (J14-3), J06-28-4-6 (J4-6), J0630-3-3 (J3-3), J06-30-3-6 (J3-6), M4-4] and a moderately resistant line (‘Old Blush’) with four susceptible roses [‘Little Chief’, RF, SC, and VS] to create F1 populations segregating for black spot resistance (Table 1). The ploidy level of the parental lines was determined by chromosome counts. All the resistant lines have black spot resistance derived from RW. These were developed by crossing RW with either ‘Old Blush’ or ‘Ducher’ and then selecting among the F2 and F3 populations created by open pollination for high black spot resistance, recurrent bloom, and a high number of flowers. The moderately resistant and susceptible parents are commercial roses with excellent ornamental characteristics. All populations were maintained as either the original seedlings or clonal cuttings from the original seedling in 1-gal pots as previously described. At the age of 2 months, the plants were pruned back to synchronize shoot development to obtain shoots of similar physiological stage for inoculation. The pruning procedure was applied each time after collecting leaf samples. Inoculation and data collection. Race 8 (also known as ACT) of D. rosae, originally collected in Brenham, TX, (Whitaker et al., 2010b) was selected as the pathogen race to use for the evaluation of partial resistance because the parents in this project showed partial resistance but not complete resistance to race 8. Thus, they do not contain Rdr3 vertical resistance gene to race 8. Conidia of race 8 of D. rosae (supplied by the Hokanson laboratory at the University of Minnesota) was collected by washing the infected leaves of ‘Cl. Pinkie’ with distilled (DI) water. The concentration of the conidia was adjusted to 1 · 10 conidia/mL with a hemocytometer (Whitaker and Hokanson, 2009a; Whitaker et al., 2010a). Detached leaf assay. From each plant in an individual pot, up to seven fully expanded young compound leaves (fourth to sixth nodes from apex of each shoot) were collected for each replication. After washing with DI (distilled) water for 10 s on each side, the seven leaves were placed on wet paper towels in a transparent plastic container (152 · 140 · 59mm). The conidia suspension (0.75 mL of 1 · 10 conidia/mL) was sprayed onto the leaves from a distance of 20–30 cm to generate fine mist, left for 48 h, and then residual water was removed by blotting. The droplet method was not used because it tended to cause leaf degradation. DI water was added to the paper towel without direct contact with the leaves to maintain the humidity in the boxes at 100%. The inoculated leaves were then maintained in the laboratory ( 25 C and 10-h photoperiod) for 14–16 d and then inspected for the incidence of acervuli under the dissecting microscope. The LAS (percentage of leaf area with symptoms) and LL (diameter of the largest individual lesion) data were collected. Although other components such as incubation period, number of lesions, and sporulation capacity could all be used to characterize the disease resistance ability, LAS and LL were chosen here due to their high linear correlations and low-labor requirements which make them appropriate for massive screening of breeding materials (Xue and Davidson, 1998). All evaluations for the rose genotype comparison and the seedling evaluations were repeated three times. Whole plant inoculation. Three plants of each genotype with branches of a similar size were selected and sprayed with a conidial suspension (1 · 10 conidia/mL) until the leaf surface was completely wet. The wet tissue was then covered with a plastic bag that was sealed around the stem for 1 week. Additional DI water was sprayed into the bags for highhumidity (100%) maintenance. The inoculated plants were then maintained in the laboratory ( 25 C, 10-h photoperiod with a humidifier) under cool white florescence lights (2150 lumens, Sylvania Fluorescent Light Bulb SL21371; Sylvania, Wilmington, MA). Four weeks after inoculation, acervuli incidence was checked under the dissecting microscope. LAS and LL data under WPI were collected on various leaves since the leaf size varied among breeding lines for similar biomass. Each plant generated one data point. The entire experiment was repeated three times. Partial resistance was measured by the diameter of the largest individual lesion (LL) and the percentage of LAS. The visual rating score of the leaf area with lesions was as Table 1. Parents, parental black spot resistance, and population size of the diploid rose populations used in the genetic study. Female parent Pollen parent Population size J14-3 (HR) SC (S) 57 SC (S) J14-3 (HR) 58 J14-3 (HR) LC (S) 140 J14-3 (HR) RF (S) 130 J14-3 (HR) VS (S) 93 VS (S) J14-3 (HR) 12 M4-4 (HR) SC (S) 26 SC (S) M4-4 (HR) 118 M4-4 (HR) VS (S) 10 J4-6 (HR) RF (S) 97 SC (S) J4-6 (HR) 23 OB (MR) J3-6 (HR) 112 OB (MR) M4-4 (HR) 54 OB (MR) RF (S) 158 J3-3 (HR) RF (S) 38 J14-3 = J06-20-14-3; J4-6 = J06-28-4-6; J3-3 = J06-30-3-3; J3-6 = J06-30-3-6; OB = ‘Old Blush’; LC = ‘Little Chief’; RF = ‘Red Fairy’; SC = ‘Sweet Chariot’; VS = ‘Vineyard Song’. S = susceptible; MR = medium resistance; HR = high resistance. Fig. 1. (A) Spores bearing acervuli on infected leaf surface of ‘Cl. Pinkie’. (B) Diagrammatic representation of leaf area with symptoms of black spot disease at 1%, 5%, 10%, 25%, 50%, or 75% in detached leaf assay. 50 HORTSCIENCE VOL. 52(1) JANUARY 2017 follows: 1 for 10%, 2 for 20%, 3 for 30%, 4 for 40%, and 5 for 50% and above for the screening of the parents that had diverse genetic backgrounds. For phenotyping the progenies of hybrid populations with similar genetic background, LAS scores were categorized as 1%, 5%, 10%, 25%, 50%, or 75% and above to detect the smaller differences of resistance among them (Fig. 1). Statistical analysis. All the boxes in DLA revealed successful inoculation with infected leaves although occasionally not all leaves showed symptoms or were severely degraded. Data were only collected on the leaves that developed symptoms and were not severely degraded to ensure the data quality. The mean LAS and LL was calculated for each box and the single value was used in further analysis. The statistical analysis was conducted by using JMP software, Version 10, and SAS software 9.3 (SAS Institute Inc., Cary, NC, 1989–2010). The normality and skewness of the population data (original and square root transformed) was analyzed by Kolmogorov–Smirnov (K–S) test (Razali and Wah, 2011), which indicated that a square root transformation improved normality and decreased skewness. Linear correlation of LL and LAS was estimated by Pearson correlation method. Genetic variances were calculated using the restricted maximum likelihood (REML) method, assuming all factors from this unbalanced design as random effects (Dieters et al., 1995; Holland et al., 2003; Littell et al., 1996). Variances of parents were considered as additive variance (VA) and variances of progeny (VP) were considered as nonadditive variance (VD) (Connor et al., 2005). Narrowsense heritability (h) was estimated by the genetic variance from the REML model, where h = VA/VP (Hallauer et al., 2010). Narrow-sense heritability was also estimated by an offspring–midparent regression (Connor et al., 2005), where h = b = cov(O, MP)/cov(MP) (Falconer and Mackay, 1996), i.e., the slope of the regression is then the estimation of heritability, with R indicating the fitness of the regression. wholesale U.S. shrub market (American-Hort, 2014). Compared with the rose market in the United States 35 years ago, the production of garden roses has decreased (Byrne et al., 2010) from 40 million roses down to 12 million field-grown and 15-18 million pot-grown rose bushes in 2012 (Hutton, 2012). This is thought to be because many rose cultivars have low tolerance to disease and abiotic stress (Waliczek et al., 2015). In response to a consumer demand for low-maintenance roses, many rose breeding programs are working toward developing cultivars resistant to the common rose diseases (Byrne, 2015;Debener and Byrne, 2014). Black spot, the most important disease affecting garden roses globally, is caused by the fungus D. rosae Wolf (Marssonina rosae anamorph) (Nauta and Spooner, 2000). The typical symptoms of this disease include dark rounded spots with a feathery edge on the adaxial side of the leaves while the abaxial epidermis remains unaffected. The disease can lead to the development of chlorosis around the lesion and eventually defoliation (Blechert and Debener, 2005;Gachomo et al., 2006;Horst and Cloyd, 2007). Eleven unique races of D. rosae have been identified among the isolates obtained from North America and Europe (Whitaker et al., 2010b). Two types of disease resistance to black spot have been characterized in roses. Vertical or complete resistance, which blocks sporulation and severely restricts the mycelial growth of the pathogen, is usually controlled by major dominant genes (Rdr or Rosa disease resistance genes) (von Malek and Debener, 1998;Whitaker et al., 2010a;Yokoya et al., 2000). In rose, the dominant resistance genes are pathogen race-specific, indicating a gene-for-gene interaction pattern (von Malek and Debener, 1998). Partial or horizontal resistance that appears to be non-race-specific has also been identified in roses (Shupert, 2005;Hokanson, 2009a, 2009b;Xue and Davidson, 1998). This resistance does not prevent infection of the pathogen, but rather delays disease development and results in reduced lesion size, reduced sporulation, and/or delayed infection after inoculation (Parlevliet, 1981;Hokanson, 2009a, 2009b;Xue and Davidson, 1998). Compared with complete resistance, partial resistance is generally more durable over the range of pathogenic races (Noack, 2003). The ideal disease-resistant genotype should have both highly effective and long-lasting resistance to a broad spectrum of pathogenic races (Blechert and Debener, 2005), which can be achieved by pyramiding dominant complete resistance genes, obtaining strong partial resistance or by combining both types of resistances. Black spot resistance is commonly evaluated in field trials in different geographic regions to expose the rose to a greater number of pathogenic races. These trials typically last 2-3 years to ensure sufficient disease pressure to properly assess the resistance of the plants (Carlson-Nilsson, 2002;Debener and Byrne, 2014;Noack, 2003;Shupert, 2005). The laboratory-based DLA is a tool for observing disease development efficiently under uniform and well-controlled environmental conditions and inoculum levels (Hattendorf et al., 2004;von Malek and Debener, 1998;Hokanson, 2009a, 2009b). Because single-conidial isolates are used in laboratory screening, the combination of compatible and incompatible interactions that are caused by various races in nature can be avoided (Blechert and Debener, 2005). However, as DLA only allows one cycle of disease development, the differences among genotypes might not be as accentuated as compared with a field trial in which multiple cycles of pathogen development are common (Xue and Davidson, 1998). Other factors such as the physical status of the host plant, degradation of the leaves, missing observations on leaf abscission/defoliation in DLA, and low or nonuniform inoculation levels in field assessment could all cause low correlation between these two methods of phenotyping (Johansson et al., 1992;Palmer et al., 1966;Zlesak et al., 2010). The objectives of this research were to 1) evaluate two methods (DLA and WPI) of artificial inoculation for black spot disease evaluation and characterize rose genotypes for black spot resistance and 2) characterize partial black spot disease resistance derived from RW in laboratory tests in diploid populations to estimate the components of genetic variances, and the heritability of this partial resistance. Fifteen diploid populations were generated in an incomplete diallel mating design by crossing five black spot-resistant lines [J06-20-14-3 (J14-3), J06-28-4-6 (J4-6), J06-30-3-3 (J3-3), J06-30-3-6 (J3-6), M4-4] and a moderately resistant line ('Old Blush') with four susceptible roses ['Little Chief', RF, SC, and VS] to create F 1 populations segregating for black spot resistance (Table 1). The ploidy level of the parental lines was determined by chromosome counts. All the resistant lines have black spot resistance derived from RW. These were developed by crossing RW with either 'Old Blush' or 'Ducher' and then selecting among the F 2 and F 3 populations created by open pollination for high black spot resistance, recurrent bloom, and a high number of flowers. The moderately resistant and susceptible parents are commercial roses with excellent ornamental characteristics. All populations were maintained as either the original seedlings or clonal cuttings from the original seedling in 1-gal pots as previously described. At the age of 2 months, the plants were pruned back to synchronize shoot development to obtain shoots of similar physiological stage for inoculation. The pruning procedure was applied each time after collecting leaf samples. Inoculation and data collection. Race 8 (also known as ACT) of D. rosae, originally collected in Brenham, TX, (Whitaker et al., 2010b) was selected as the pathogen race to use for the evaluation of partial resistance because the parents in this project showed partial resistance but not complete resistance to race 8. Thus, they do not contain Rdr3 vertical resistance gene to race 8. Conidia of race 8 of D. rosae (supplied by the Hokanson laboratory at the University of Minnesota) was collected by washing the infected leaves of 'Cl. Pinkie' with distilled (DI) water. The concentration of the conidia was adjusted to 1 · 10 5 conidia/mL with a hemocytometer (Whitaker and Hokanson, 2009a;Whitaker et al., 2010a). Detached leaf assay. From each plant in an individual pot, up to seven fully expanded young compound leaves (fourth to sixth nodes from apex of each shoot) were collected for each replication. After washing with DI (distilled) water for 10 s on each side, the seven leaves were placed on wet paper towels in a transparent plastic container (152 · 140 · 59 mm). The conidia suspension (0.75 mL of 1 · 10 5 conidia/mL) was sprayed onto the leaves from a distance of 20-30 cm to generate fine mist, left for 48 h, and then residual water was removed by blotting. The droplet method was not used because it tended to cause leaf degradation. DI water was added to the paper towel without direct contact with the leaves to maintain the humidity in the boxes at 100%. The inoculated leaves were then maintained in the laboratory (25°C and 10-h photoperiod) for 14-16 d and then inspected for the incidence of acervuli under the dissecting microscope. The LAS (percentage of leaf area with symptoms) and LL (diameter of the largest individual lesion) data were collected. Although other components such as incubation period, number of lesions, and sporulation capacity could all be used to characterize the disease resistance ability, LAS and LL were chosen here due to their high linear correlations and low-labor requirements which make them appropriate for massive screening of breeding materials (Xue and Davidson, 1998). All evaluations for the rose genotype comparison and the seedling evaluations were repeated three times. Whole plant inoculation. Three plants of each genotype with branches of a similar size were selected and sprayed with a conidial suspension (1 · 10 5 conidia/mL) until the leaf surface was completely wet. The wet tissue was then covered with a plastic bag that was sealed around the stem for 1 week. Additional DI water was sprayed into the bags for highhumidity (100%) maintenance. The inoculated plants were then maintained in the laboratory (25°C, 10-h photoperiod with a humidifier) under cool white florescence lights (2150 lumens, Sylvania Fluorescent Light Bulb SL21371; Sylvania, Wilmington, MA). Four weeks after inoculation, acervuli incidence was checked under the dissecting microscope. LAS and LL data under WPI were collected on various leaves since the leaf size varied among breeding lines for similar biomass. Each plant generated one data point. The entire experiment was repeated three times. Partial resistance was measured by the diameter of the largest individual lesion (LL) and the percentage of LAS. The visual rating score of the leaf area with lesions was as follows: 1 for 10%, 2 for 20%, 3 for 30%, 4 for 40%, and 5 for 50% and above for the screening of the parents that had diverse genetic backgrounds. For phenotyping the progenies of hybrid populations with similar genetic background, LAS scores were categorized as 1%, 5%, 10%, 25%, 50%, or 75% and above to detect the smaller differences of resistance among them (Fig. 1). Statistical analysis. All the boxes in DLA revealed successful inoculation with infected leaves although occasionally not all leaves showed symptoms or were severely degraded. Data were only collected on the leaves that developed symptoms and were not severely degraded to ensure the data quality. The mean LAS and LL was calculated for each box and the single value was used in further analysis. The statistical analysis was conducted by using JMP software, Version 10, and SAS software 9.3 (SAS Institute Inc., Cary, NC, 1989. The normality and skewness of the population data (original and square root transformed) was analyzed by Kolmogorov-Smirnov (K-S) test (Razali and Wah, 2011), which indicated that a square root transformation improved normality and decreased skewness. Linear correlation of LL and LAS was estimated by Pearson correlation method. Genetic variances were calculated using the restricted maximum likelihood (REML) method, assuming all factors from this unbalanced design as random effects (Dieters et al., 1995;Holland et al., 2003;Littell et al., 1996). Variances of parents were considered as additive variance (V A ) and variances of progeny (V P ) were considered as nonadditive variance (V D ) (Connor et al., 2005). Narrowsense heritability (h 2 ) was estimated by the genetic variance from the REML model, where h 2 = V A /V P (Hallauer et al., 2010). Narrow-sense heritability was also estimated by an offspring-midparent regression (Connor et al., 2005), where h 2 = b = cov(O, MP)/cov(MP) (Falconer and Mackay, 1996), i.e., the slope of the regression is then the estimation of heritability, with R 2 indicating the fitness of the regression. Results Phenotyping of parental materials. Sporebearing acervuli were observed on all parental genotypes whether using the DLA or WPI method, indicating that complete resistance to race 8 of D. rosae did not exist among the selected rose genotypes. Using LL and LAS data, the most resistant genotypes as determined by field observations (RW, M4-4, and J4-6) were clearly distinguishable from the roses rated as most susceptible to D. rosae (RF, SC, Cal, VS, and OH) ( Table 2). The best resolution among rose genotypes was with the LAS data, which was also able to separate other field-resistant roses (91/100-5, DD, J3-6) from the susceptible genotypes. LAS data indicated that the highest rating is 2.46 and 2.44 obtained from OH from DLA and WPI, respectively, either due to the resistance ability from the breeding lines selected for this study or the pathogenesis ability of race 8. Although various resistant abilities could be revealed by different measurement components, the trending of susceptibility indicated by artificial inoculation is similar to what has been observed in the field. Since partial resistance is a continuous trait, clearly defining a germplasm as resistant or susceptible material might not be practical. The two measures of resistance, LAS and LL, are well correlated (R = 0.91 at P < 0.0001; Fig. 2). Genotypes with a higher percentage of LAS being covered with lesions showed longer LL, indicating either of these two traits could be used as an indicator of the host plant response to the pathogen. When using the WPI to quantify the black spot resistance of the genotypes, it was found that the rose genotypes with higher resistance generally had lower LAS, LL, and number of fallen leaves (NF) when compared with the most susceptible rose genotypes, but these groups were not consistently different (Table 2). This would suggest that the DLA approach is the better method for quantifying the relative partial resistance of rose to black spot. The various measures of black spot, LAS and LL, are generally correlated between DLA and WPI (R ranging from 0.46 to 0.58). With the WPI approach, although NF was correlated with LL, LL and LAS were not correlated. LAS and LL data from DLA were well correlated ( Table 3). The significant correlation between DLA and WPI, revealed by this study (R ranging from 0.56 to 0.58), is slightly weaker than the research conducted by Whitaker and Hokanson (2009a), which may be due to the different genetic background of the materials used for screening or inoculation methods such as growth conditions and the races of D. rosae that were selected. Disease distribution of diploid populations. Based on the results of K-S normality test, the LL and LAS data normality improved and skewness generally decreased after a square root transformation. Thus, all subsequent statistical analyses were done with the transformed data, but it should be noted that the conclusions reached with the untransformed data and transformed data were not different. Correlations among resistance assessments. The correlation of individual progenies' partial resistance to black spot race 8 measured by LAS and LL (squareroot-transformed data) from the detached leaf tests is 0.3 (P < 0.0001) (Fig. 3). The correlation of these two components was much higher (R = 0.9) when estimating among resistant and susceptible parental materials, probably due to the smaller range of resistance among the seedlings vs. parents as measured by LL (0.5-3.0 mm vs. 0.1-7.14 mm). Genetic variation and estimation of heritability of disease assessments using the DLA. In this incomplete diallel mating design, narrow-sense heritability (additive variance/phenotypic variance) was estimated to be 0.28 and 0.43 for LAS and LL, respectively, indicating this partial resistance trait is moderately heritable. The parental variance for LAS and LL account for 24% and 34% of the variance, while the progeny variance for LAS and LL accounts for 61% and 45% of total genetic variance, respectively, indicating important nonadditive genetic effects. In contrast, the narrow-sense heritability estimation using the offspring-midparent regression approach was 0.86 and 0.74 for LAS and LL, respectively. The fitness of the regressions (R 2 ) of LAS and LL was calculated as 0.47 and 0.43, respectively, indicating a fairly good estimation of the midparent and offspring performances (Fig. 4). Discussion Artificial inoculation on parental germplasm. From this project, several cultivars (RF, Cal, SC, VS, and OH) were rated as very susceptible to black spot. Interestingly, the breeding line J3-6, which is derived from the wild species RW and has a high level of partial resistance to black spot, had greater defoliation under WPI than the other resistant lines (Table 2). In contrast, Cal, a susceptible material based on field observation, showed no defoliation under WPI. It is possible that different resistant mechanisms occurred in the host plants. Artificial inoculation results indicated that DLA could distinguish the performance of the genotypes better than WPI, and the LL and LAS of DLA were well correlated as previously reported by Xue and Davidson (1998). Our results also confirm previous studies (Whitaker and Hokanson, 2009b) that with the DLA approach, it is easier to create a uniform humid environment and it requires less input of time and facilities as compared with the WPI. The DLA is more appropriate for the phenotyping of large populations and cultivar collections. However, as LAS and LL data generated from WPI were well correlated with LAS and LL data generated from DLA (Table 3), WPI could be used as a complementary characterization method to DLA for those genotypes whose leaves degraded easily. Genetic variances calculated from the mixed model (both LAS and LL) indicated that the additive variances explained 24% to 34% of the total variances, which is lower than that explained by nonadditive variances (45% to 61%) (Table 4). In contrast, the midparent-progeny mean regression indicated that both measures were mainly additive in inheritance with heritability estimates of 0.74-0.86. The normal breeding approach for rose, a clonally propagated crop, is recurrent mass selection. Thus, selection is based on the phenotype of the individual seedlings with the best phenotypes being used as parents for the next generation. This is an efficient approach for traits with high narrow-sense heritability. The narrow-sense heritability estimates for LL and LAS ranged by estimation approach REML (h 2 = 0.28-0.43) and midparent regression (h 2 = 0.74-0.86), indicating this approach should be moderately to highly effective. Nevertheless, if the nonadditive genetic variance is as large as the additive genetic variance as indicated by the REML analysis, the progeny performance of specific parents should also be factored into the choice of parents in population creation. From a complete factorial mating design of partially resistant and susceptible roses conducted by Whitaker and Hokanson (2009a), within-family variances are much lower than that of between-family variances. Therefore, selection for certain families (generated from specific parental combinations) followed by backcrossing to the parents with more advanced ornamental traits was suggested for future breeding as an efficient selection/breeding approach. Different results of narrow-sense heritability estimated from genetic variances and offspring-midparent regression might due to the structure of hybrid populations as the incomplete diallel mating design used in this study reduces the power of estimating genetic variances. In either case, the data indicate that it should be possible to develop a rose with stable and high resistance to black spot based on partial resistance derived from RW. = additive variances based on variances of parents; s 2 D = nonadditive variances based on variance of progeny. y Percent of total variances = percentage of total genetic variances caused by s 2 A , s 2 D , and s 2 P . x Narrow-sense heritability = ratio of additive genetic variance to total phenotypic variance; h 2 = s 2 A /s 2 P .	2019-04-02T13:04:23.884Z	2017-01-01T00:00:00.000	{ "year": 2017, "sha1": "c903f1606af14285d7574554951bae4106216900", "oa_license": null, "oa_url": "https://journals.ashs.org/downloadpdf/journals/hortsci/52/1/article-p49.pdf", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "567778fc5d28f4aee8e17df14f01239783a1747f", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Biology" ] }
243359708	pes2o/s2orc	v3-fos-license	A meta-analysis on the curative effect and safety of high-dose flurbiprofen axetil for pain management after general surgery Background To compare the curative efficiency or tolerability of flurbiprofen axetil(FA) in a high dose with that in a standard dose for postoperative pain of general surgery. Methods Relevant RCTs were retrieved from PubMed, Ovid, EMBASE, the Cochrane library, CBM, and CNKI from their inceptions to July 2019. The included studies were selected according to eligibility criteria. The study design, participant characteristics, interventions, and outcomes were abstracted after assessingmethodological quality of the trials. All data were analyzed by Review Manager 5.3. 10 studies were identified, which compared the curative effect or tolerability of FA between high and standard dose group. 500 patients were involved in this meta-analysis, with 250 patients in high dose group and 250 patients in standard dose group, respectively. Pooled analysis of VAS scores at 1, 2, 4, 6, 8, 12, and 24h showed that VAS scores at 1h(P <0.00001), at 2h (P=0.003), at 4h (P = 0.0007), at 6h (P=0.0002), at 8h (P= 0.0002), at 12h (P = 0.0001), and at 24h (P = 0.0004) in the high dose group were significantly lower than that in the standard group. Pooled analysis of BCS scores at 1, 2, 4, 6, 8, 12, and 24h showed that BCS scores at 1h (P < 0.00001), at 6h (P < 0.00001), at 12h(P=0.03), and at 24h (P=0.01) in the high dose group were significantly higher than that in the standard group. Pooled analysis showed that there was no difference in the incidence of adverse events or administrating rate of analgesics after FA treatment between high and standard dose groups (P>0.05). 3 effective than that with standard dose in postoperative pain control after general operation,while the incidence of adverse effects with high or standard dose showed no significant difference. Background The treatment of pain is an important concern both for patients and anesthetists. Postoperative pain is caused by direct surgical trauma, and its intensity and range are usually positively correlated with the extent of surgery. Postoperative pain not only affects the living quality of the patients, but also increases the incidence of complications after surgery, and finally delays the recovery of physical function. Therefore, effective analgesic therapy becomes an integral part of treating a 'perioperative disease' As a nonselective cyclooxygenase (COX) inhibitor, flurbiprofen axetil (FA) owns a high affinity to the site of surgical incision and inflammatory tissues for being incorporated in lipid micro-balloon spheres. FA exerts their analgesic effect through inhibiting the biosynthesis of prostaglandins, restraining sensitization of the peripheral and central nervous systems [1]. FA, a nonsteroidal anti-inlammatory drug (NSAID), is routinely applied in the control of postoperative pain in clinic. Early published studies have showed that preoperative administration of FA reduces postoperative pain in patients undergoing general surgeries, such as thyroidectomy,cholecystectomy, mastectomy, etc. [2][3][4] However, there is no clear definition about the optimal analgesic dose of FA in the management of postoperative pain. Frequently, the recommended dose of FA is 50 mg/injection [5]. Recently, some data from several research showed that FA with larger than the recommended dose were more effective in treating postoperative pain [6][7][8][9][10][11][12][13]. However, incidence of side effects might increase with higher dose, ingestion of larger doses is limited. Currently, the maximal dosage of FA is 250 mg [14]/day but the optimal dose for a single injection remains unclear when the total amout of 24h is below 250mg. 4 Administrating with FA in doses larger than the recommended 50mg/injection hasn't been formally assessed for effectiveness and safety. Toward this end, we set out to compare the efficacy and safety of high versus standard doses of FA in patients undergoing general surgeries. Search for eligible studies We identified relevant studies through searching electronic databases of PubMed, Ovid, EMBASE, the Cochrane library, CBM and CNKI using the following keywords: ''flurbiprofenaxetil'', "dose", "surgery", and "pain" (from their inceptions to July 2019). The reference lists of included articles and relevant reviews were searched manually to find other potentially eligible studies. Inclusion and exclusion criteria Articles were selected on the basis of the following criteria: 1) Randomized controlled trials (RCTs); 2) Comparing the efficacy and tolerability of FA with a high single dose (≥1.25mg/kg) to that with a standard single dose (50mg) for postoperative pain; 3) Patients who experienced postoperative acute pain after general surgery (including hepatobiliary and pancreatic surgery, gastrointestinal surgery, thyroid surgery, anorectal surgery, breast surgery, and vascular surgery); 4) The clinical outcomes of pain intensity were evaluated. Exclusion criteria were as follows: 1) Studies on breakthrough pain; 2) Studies compared the effiency of FA adiminstrated at different operation period; 3) Studies combined FA preparations; 4) Letters, case reports, editorials or reviews; 5) Studies with incomplete raw data. Data extraction For eligible studies, two reviewers independently extracted data from full-text articles using a predefined form. The following information were collected: First author, year of publication, intervention methods, number of cases, gender ratio, mean age, outcome measures (Visual Analogue Scale (VAS), Bruggrmanncomfort scale (BCS) ), complications and adverse reaction rate and information relevant to trial quality allocation concealment, blinding, ect . Uncertainty in the determination of eligibility was resolved by discussion with the other reviewer. If there were studies which has only one subgroup of the participants met the inclusion criteria , we would only extract data on this subgroup to perform the current meta-analysis. Study Quality Two reviewers independently evaluated the risk of bias using the Cochrane Collaboration tool [15]. The authors estimated the following domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. Based on the information extracted from included studies, each domain was assigned as a value of ''high risk,'' ''unclear risk,'' or ''low risk.'' Any disagreements between searchers concerning the eligibility of a trial were resolved by consulting a third reviewer. Statistical Analysis Adverse reaction rate and administration rate of analgetic drug were binary outcome data with odds ratio (OR) and 95% confidence intervals (CIs). VAS and BCS scores were continuous outcome data with mean difference (MD) and respective 95% . We used the program of the Cochrane Collaboration (Review Manager 5.3) to calculate the summary statistic for each component study. The random effect model or fixed effect model was used for outcomes analyse of continuous or dichotomized variables. When there was no difference between the findings derived from the 2 models, the fixed-effect model was used for the results. This was expected to happen in the absence of significant statistical 6 heterogeneity. Funnel plots were employed to assess the possibility of publication bias. These plots showed the intervention effect from each study against the respective standard error. A symmetrical plot reveals no bias, and any asymmetry of the plot would suggest publication bias. The sensitivity analysis was performed to test the strength and robustness of pooled results by sequential omission of individual studies. Results Search results and characteristic of included studies A total of 370 relevant titles were identified through searching databases. Of these, 350 were excluded after reviewing abstracts and titles for being on an unrelated topic; not postoperativepain;or not RCTs. Finally, 19 full text articles were collected, wherein ten studies met the inclusion criteria [7-13 16-18] (Figure 1). The characteristics of the 10 studies were summarized in Table 1. 500 patients were included in this meta-analysis, including 250 patients in high dose group and 250 patients in standard dose group, respectively. Assessing risk of bias The detail of the risk-of-bias assessment of included studies was summarized in Figure 2 and Figure 3. Seven eligible studies were incorporated for risk bias assessment. In terms of the adequate randomization sequence, all studies were assessed as low risk. However, many relative information in the studies wasn't available, such as allocation concealment and blinding of participants and personnel, blinding of outcome assessment. Nevertheless, the overall methodological quality was generally fair. 6). Administrating rate of analgetic drug 8 The administrating rate of analgeic drug was available in 3 studies. Pooled analysis showed that there was no difference in administrating rate of analgetic drug after FA treatment between high and standard dose groups (OR, 0.57; 95% CI, [0.20, 1.65]; P=0.30) (Figure 7). Discussion Up to now, the optimal dosage of FA for preventive analgesia remained uncertain, but 50 mg was chosen as the standard dose in clinical study. Some studies have evaluated postoperative analgesic effect of FA in different doses and presented a dose-related effect [19] [6] [7][8][9][10][11][12][13]. In this meta-analysis, we combined all data series about FA in different doses for postoperative control after general surgeries to determine the optimal dosage of FA. Compared with a single study, the study of larger sample size was more likely to get precise conclusions. As far as we know, this study was the first meta-analysis on the topic. In the meta-analysis, we found that VAS scores in the high dose group were significantly lower than that in the standard group at 1, 2, 4, 6, 8, 12, and 24h, and BCS scores in the high dose group were significantly higher than that in the standard group at 1, 6, 12, and 24h after FA treatment. No significant difference in adverse effects incidence or administrating rate of analgetic drug was found between high and standard dose groups . VAS and BCS are frequently used to evaluate the intensity of pain. Our study pointed out that the postoperative analgesic effect of FA in high dose was superior to that of the standard dose. Mikawa K et al. performed a prospective, randomised, controlled trial of 90 children to reveal that preoperative treatment using 1 mg/kg flurbiprofen was more effective than that of 0.5 mg/kg for postoperative pain after paediatric strabismus surgery [19]. In a systematic review, oral flurbiprofen 50 mg was used in acute postoperative pain.The 9 result showed that the number needed to treat (NNT) to benefit for at least 50% pain relief over 4 to 6 hours was 2.7 (2.3 to 3.3) , and for 100 mg it was 2.5 (2.0 to 3.1) compared with placebo.The incidence of adverse effects in placebo group was not significantly different from that of flurbiprofen 50 mg or 100 mg groups. The results were compatible with a dose response, but the differences weren't significant [20]. Schachtel BP also proposed the dose-response relationship of flurbiprofen lozenges for treating sore throat in 3 dosages (2.5, 5.0, and 12.5 mg) [21]. In a recent study by Zhao X et al , the analgesic effect of different doses of flurbiprofen axetil was also analyzed for postoperative analgesia, and the results showed that VAS score and postoperative pain incidence in group with the dose of 150 mg or 200 mg was lower than those in the dose of 100 mg. Group with the dose of 150 mg had better analgesic effect and lower incidence of adverse reactions [22]. Conclusions In the meta-analysis, we demonstrated that the postoperative analgesia in the high dose provided by 1.25-2 mg/kg FA was superior to that of standard dose, and the side-effects were comparable. Limitation Our study has some limitations. Firstly, the quality of included studies was low. Secondly, publication bias was existed, for all included studies were from China. Further clinical trials with high quality were needed to restate the significant dose response and to confirm that using a higher dose provides additional benefit. Despite these weaknesses, this systematic review can still provide some value for clinical practice. Meta-analysis of VAS score between the high dose group and standard group 19 Figure 5 Meta-analysis of BCS score between the high dose group and standard group 20 Figure 6 Meta-analysis of adverse reaction rate between the high dose group and standard group Figure 7 Meta-analysis of the administration rate of analgetic drug rate between the high dose group and standard group Supplementary Files This is a list of supplementary files associated with the primary manuscript. Click to download. PRISMA guidelines.pdf	2019-09-17T00:49:39.195Z	2019-08-15T00:00:00.000	{ "year": 2019, "sha1": "c1f0175a08ddedc67b570e9e71b29abf03688645", "oa_license": "CCBY", "oa_url": "https://www.researchsquare.com/article/rs-3761/v1.pdf", "oa_status": "GREEN", "pdf_src": "Adhoc", "pdf_hash": "20da0d0db3a01b96a55ee80f76b4c4930e8976ac", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [] }
54169945	pes2o/s2orc	v3-fos-license	Theory and Simulation of Spin Transport in Antiferromagnetic Semiconductors: Application to MnTe We study in this paper the parallel spin current in an antiferromagnetic semiconductor thin film where we take into account the interaction between itinerant spins and lattice spins. The spin model is an anisotropic Heisenberg model. We use here the Boltzmann's equation with numerical data on cluster distribution obtained by Monte Carlo simulations and cluster-construction algorithms. We study the cases of degenerate and non-degenerate semiconductors. The spin resistivity in both cases is shown to depend on the temperature with a broad maximum at the transition temperature of the lattice spin system. The shape of the maximum depends on the spin anisotropy and on the magnetic field. It shows however no sharp peak in contrast to ferromagnetic materials. Our method is applied to MnTe. Comparison to experimental data is given. I. INTRODUCTION The behavior of the spin resistivity ρ as a function of temperature (T ) has been shown and theoretically explained by many authors during the last 50 years. Among the ingredients which govern the properties of ρ, we can mention the scattering of the itinerant spins by the lattice magnons suggested by Kasuya 1 , the diffusion due to impurities 2 , and the spin-spin correlation. [3][4][5] First-principles analysis of spin-disorder resistivity of Fe and Ni has been also recently performed. 6 Experiments have been performed on many magnetic materials ranging from metals to semiconductors. These results show that the behavior of the spin resistivity depends on the material : some of them show a large peak of ρ at the magnetic transition temperature T C , 7 others show only a change of slope of ρ giving rise to a peak of the differential resistivity dρ/dT . 8,9 Very recent experiments such as those performed on ferromagnetic SrRuO 3 thin films 10 have showed that in magnetic diluted semiconductors the shape of the resistivity versus T depends on the interaction between the itinerant spins and localized magnetic impurities which is characterized by a Anderson localization length ζ. Expressing physical quantities in terms of ζ around impurities, they calculated ρ and showed that its peak height depends indeed on this localization length. In our previous work [17][18][19] we have studied the spin current in ferromagnetic thin films. The behavior of the spin resistivity as a function of T has been shown and explained as an effect of magnetic domains formed in the proximity of the phase transition point. This new concept has an advantage over the use of the spin-spin correlation since the distribution of clusters is more easily calculated using Monte Carlo simulations. Although the formation of spin clusters and their sizes are a consequence of spin-spin correlation, the direct access in numerical calculations to the structure of clusters allows us to study complicated systems such as thin films, systems with impurities, systems with high degree of instability etc. On the other hand, the correlation functions are very difficult to calculate. Moreover, as will be shown in this paper, the correlation function cannot be used to explain the behavior of the spin resistivity in antiferromagnets where very few theoretical investigations have been carried out. One of these is the work by Suezaki and Mori 20 which simply predicted that the behavior of the spin resistivity in antiferromagnets is that in ferromagnets if the correlation is short-ranged. It means that correlation should be limited to "selected nearestneighbors". Such an explanation is obviously not satisfactory in particular when the sign of the correlation function between antiparallel spin pairs are taken into account. In a work with a model suitable for magnetic semiconductors, Haas has shown that the resistivity ρ in antiferromagnets is quite different from that of ferromagnets. 21 In particular, he found that while ferromagnets show a peak of ρ at the magnetic transition of the lattice spins, antiferromagnets do not have such a peak. We will demonstrate that all these effects can be interpreted in terms of clusters used in our model. In this paper, we introduce a simple model which takes into account the interaction between itinerant spins and localized lattice spins. This is similar to the s − d model 21 . The lattice spins interact with each other via antiferromagnetic interactions. The model will be studied here by a combination of Monte Carlo simulation and the Boltzmann's equation. As will be discussed below, such a model corresponds to antiferromagnetic semiconductors such as MnTe. An application is made for this compound in the present work. The paper is organized as follows. In section II, we show and discuss our general model and its application to the antiferromagnetic case using the Boltzmann's equation formulated in terms of clusters. We also describe here our Monte Carlo simulations to obtain the distributions of sizes and number of clusters as functions of T which will be used to solve the Boltzmann's equation. Results on the effects of Ising-like anisotropy and magnetic field as well as an application to the case of MnTe is shown in section III. Concluding remarks are given in section IV. II. THEORY Let us recall briefly principal theoretical models for magnetic resistivity ρ. In magnetic systems, de Gennes and Friedel 3 have suggested that the magnetic resistivity is proportional to the spin-spin correlation. As a consequence, in ferromagnetically ordered systems, ρ shows a divergence at the transition temperature T C , similar to the susceptibility. However, in order to explain the finite cusp of ρ experimentally observed in some experiments, Fisher and Langer 4 suggested to take into account only short-range correlations in the de Gennes-Friedel's theory. Kataoka 5 has followed the same line in proposing a model where he included, in addition to a parameter describing the correlation range, some other parameters describing effects of the magnetic instability, the density of itinerant spins and the applied magnetic field. For antiferromagnetic systems, Suezaki and Mori 20 proposed a model to explain the anomalous behavior of the resistivity around the Néel temperature. They used the Kubo's formula for an s − d Hamiltonian with some approximations to connect the resistivity to the correlation function. However, it is not so easy to resolve the problem. Therefore, the form of the correlation function was just given in the molecular field approximation. They argued that just below the Néel temperature T N a long-range correlation appears giving rise to an additional magnetic potential which causes a gap. This gap affects the electron density which alters the spin resistivity but does not in their approximation interfere in the scattering mechanism. They concluded that, under some considerations, the resistivity should have a peak close to the Néel point. This behavior is observed in Cr, α − Mn and some rare earth metals. Note however that in the approximations used by Haas 21 , there is no peak predicted. So the question of the existence of a peak in antiferromagnets remains open. Following Haas, we use for semiconductors the following interaction where J( r − R n ) is the exchange interaction between an itinerant spin s at r and the lattice spin S n at the lattice site R n . In practice, the sum on lattice spins S n should be limited at some cut-off distance as will be discussed later. Haas supposed that V is weak enough to be considered as a perturbation to the lattice Hamiltonian given by Eq. (15) below. This is what we also suppose in the present paper. He applied his model to ferromagnetic doped where f 0 is the equilibrium Fermi-Dirac function, k the wave vector, e and m the electronic charge and mass, ǫ the electron energy. We next use the following relaxation-time where τ k is the relaxation time. Supposing elastic collisions, i. e. k = k ′ , and using the detailed balance we have where Ω is the system volume, w k ′ ,k the transition probability between k and k ′ . We find with Eq. (3) and Eq. (4) the following well-known expression where θ and φ are the angles formed by k ′ with k, i. e. spherical coordinates with z axis parallel to k. We use now in Eq. (5) the "Fermi golden rule" for ω k,k ′ and we obtain where J(r) is the exchange integral between an itinerant spin and a lattice spin which is given in the scattering potential, Eq. (1). One has Note that for simplicity we have supposed here that the interaction potential J(r) depends only on the relative distance r ′ = \| r − R n \|, not on the direction of r − R n . We suppose in the following a potential which exponentially decays with distance where V 0 expresses the magnitude of the interaction and ξ the averaged cluster size. After some algebra, we arrive at the following relaxation time where k f is the Fermi wave vector. As noted by Haas 21 , the mobility is inversely proportional to the susceptibility χ. So, in examining our expression and in using the following expression 28 where η(ξ) is the number of clusters of size ξ, one sees that the first term of the relaxation time is proportional to the susceptibility. The other two terms are the corrections. The mobility in the x direction is defined by We resolve the mobility µ x explicitly in the following two cases -Degenerate semiconductors . We arrive at the following mobility The resistivity is then We can check that the right-hand side has the dimension of a resistivity : [Ω][m]. -Non-degenerate semiconductors One has in this case f 0 Note that the formulation of our theory in terms of cluster number η and cluster size ξ is numerically very convenient. These quantities are easily calculated by Monte Carlo simulation for the Ising model. The method can be generalized to the case of Heisenberg spins where the calculation is more complicated as seen below. In section III A we will examine values of parameter V 0 where the Born's approximation is valid. where S i is the Heisenberg spin at the site i, i,j is performed over all nearest-neighbor (NN) spin pairs. We assume here that all interactions including those at the two surfaces are identical for simplicity : J is positive (antiferromagnetic), and A an Ising-like anisotropy which is a positive constant. When A is zero, one has the isotropic Heisenberg model and when A → ∞, one has the Ising model. The classical Heisenberg spin model is continuous, so it allows the domain walls to be less abrupt and therefore softens the behavior of the magnetic resistance. Note that for clarity of illustration, in this section (II.B) we suppose only NN interaction J. In the application to MnTe shown in section III.C, the exchange integral is distance-dependent and we shall take into account up to the third NN interaction. Hereafter, the temperature is expressed in unit of J/k B , k B being the Boltzmann's constant. A is given in unit of J. The resistivity ρ is shown in atomic units. For the whole paper, we use N x = N y = 20, and N z = 8. The finite-size effect as well as surface effects are out of the scope of the present paper. Using the Hamiltonian (15), we equilibrate the lattice at a temperature T by the standard Monte Carlo simulation. In order to analyze the spin resistivity, we should know the energy landscape seen by an itinerant spin. The energy map of an itinerant electron in the lattice is obtained as follows : at each position its energy is calculated using Eq. (8) within a cutoff at a distance D 1 = 2 in unit of the lattice constant a. The energy value is coded by a color as shown in Fig. 1 for the case A = 0.01. As seen, at very low T (T = 0.01) the energy map is periodic just as the lattice, i. e. no disorder. At T = 1, well below the Néel temperature T N ≃ 2.3, we observe an energy map which indicates the existence of many large defect clusters of high energy in the lattice. For T ≈ T N the lattice is completely disordered. The same is true for T = 2.5 above T N . We shall now calculate the number of clusters and their sizes as a function of T in order to analyze the temperature-dependent behavior of the spin current. Note that we can define a cluster distribution by each value of S z . We can therefore distinguish the amplitude of scattering : as seen below scattering is stronger for cluster with larger S z . We have used the above procedure to count the number of clusters in our simulation of an antiferromagnetic thin film. We show in Fig. 3 Lines are guides to the eye. A. Effect of Ising-like Anisotropy At this stage, it is worth to return to examine some fundamental effects of V 0 and A. It is necessary to know acceptable values of V 0 imposed by the Born's approximation. To do Lines are guides to the eye. this we must calculate the resistivity with the second order Born's approximation. The first term is due to the first order of Born's approximation and the second and third terms to corrections from the second order. We plot ρ(Born2)/ρ(Born1) versus T in Fig. 5 for different values of V 0 , ρ(Born1) and ρ(Born2) being respectively the resistivities calculated at the first and second order. We note that the larger this ratio is, the more important the corrections due to the second-order become. From Fig. 5, several remarks are in order : -The first order of Born's approximation is valid for small values of V 0 as seen in the case V 0 = 0.01 corresponding to a few meV. In this case the resistivity does not depend on T . This is understandable because with such a weak coupling to the lattice, itinerant spins do not feel the effect of the lattice spin disordering. -In the case of strong V 0 such as V 0 = 0.05, the second-order approximation should be used. Interesting enough, the resistivity is strongly affected by T with a peak corresponding to the phase transition temperature of the lattice. We examine now the effect A. Figure 6 shows the variation of the sublattice magnetization and of T N with anisotropy A. We have obtained respectively for A = 0.01, A = 1, A = 1.5 and pure Ising case the following critical temperatures T N ≃ 2.3, 4.6, 5.6 and 6.0. Note that the pure Ising case has been simulated with the pure Ising Hamiltonian, not with Eq. (15) (we cannot use A = ∞). We can easily understand that not only the spin resistivity will follow this variation of T N but also the change of A will fundamentally alter the resistivity behavior as will be seen below. The results shown in Fig. 7 indicate clearly the appearance of a peak at the transition which diminishes with increasing anisotropy. If we look at Fig. 4 which shows the average size of clusters as a function of T , we observe that the size of clusters of large S z diminishes with increasing A. We show in Fig. 8 show no peak at all 31,32 . These results are in agreement with the tendency observed here for increasing A. B. Effect of Magnetic Field We apply now a magnetic field perpendicularly to the electric field. To see the effect of the magnetic field it suffices to replace the distribution function by From this, we obtain the following equations for the contributions of up and down spins where S z is the domain-wall spin (scattering centers) and V 0 is the coefficient of the exchange integral between an itinerant spin and a lattice spin [see Eq. (8)]. Figures 9 and 10 show the resistivity for several magnetic fields. We observe a split in the resistivity for up and down spins which is larger for stronger field. Also, we see that the minority spins shows a smaller resistivity due to their smaller number. The reason is similar to the effect of A mentioned above and can be understood by examining Fig. 11 where we Lines are guides to the eye. pure intrinsic semiconductors, k f is in the gap and its position is given by the law of mass action using parabolic band approximation. In doped cases, band tails created by doped impurities can cover more or less the gap. But this system, which is disordered by doping, is not a purpose of our present study. In semiconductors valence electrons can go from the valence band to the conduction band more and more as the temperature increases. Therefore, the carrier concentration is a function of T . Our model has a number of itinerant spins which is independent of T in each simulation. However in each simulation, we can take another concentration (see Ref. 19) : the results show that the resistivity is not strongly modified, one still has the same feature, except that the stronger the concentration is the smaller the peak at T C becomes if and only if interaction between itinerant spins is taken into account. Therefore, we believe that generic effects independent of carrier concentration will remain. Of course, the correct way is to use a formula to generate the carrier concentration as a function of T and to make the simulation with the temperature-dependent concentration taking account additional scattering due to interaction between itinerant spins. Unfortunately, to obtain that formula we have to use several approximations which involve more parameters. We will try this in a future work. In the case of Cd 1−x Mn x Te, the question of the crystal structure, depending on the doping concentration x remains open. Cd 1−x Mn x Te can have one of the following structures, the so-called NiAs structure or the zinc-blend one, or a mixed phase. [33][34][35][36] The pure MnTe crystallizes in either the zinc-blend structure 37 or the hexagonal NiAs one 38 (see Fig. 12). MnTe is a well-studied p-type semiconductor with numerous applications due to its high Néel temperature. We are interested here in the case of hexagonal structure. For this case, the Néel temperature is T N = 310 K 38 . We have calculated the cluster distribution for the hexagonal MnTe using the exchange integrals taken from Ref. 38 and the other crystal parameters taken from the literature [39][40][41] . The result is shown in Fig. 13. The spin resistivity in MnTe obtained with our theoretical model is presented in Fig. 14 iv) the existence of the peak at T N = 310 K of the theoretical spin resistivity shown in To close this section, let us note that it is also possible, with some precaution, to apply our model on other families of antiferromagnetic semiconductors like CeRhIn 5 and LaFeAsO. An example of supplementary difficulties but exciting subject encountered in the latter compound is that there are two transitions in a small temperature region : a magnetic transition at 145 K and a tetragonal-orthorhombic crystallographic phase transition at 160 K. 25,26 An application to ferromagnetic semiconductors of the n-type CdCr 2 Se 4 42 is under way. IV. CONCLUSION We have shown in this paper the behavior of the magnetic resistivity ρ as a function of temperature in antiferromagnetic semiconductors. The main interaction which governs the resistivity behavior is the interaction between itinerant spins and the lattice spins. Our analysis, based on the Boltzmann's equation which uses the temperature-dependent cluster distribution obtained by MC simulation. Our result is in agreement with the theory by Haas 21 : we observe a broad maximum of ρ in the temperature region of the magnetic transition without a sharp peak observed in ferromagnetic materials. We have studied the two cases, degenerate and non-degenerate semiconductors. The non-degenerate case shows a maximum which is more pronounced than that of the degenerate case. We would like to emphasize that the shape of the maximum and its existence depend on several physical parameters such as interactions between different kinds of spins, the spin model, the crystal structure etc. In this paper we applied our theoretical model in the antiferromagnetic semiconductor MnTe. We found a good agreement with experimental data near the transition region. We note however that our model using the cluster distribution cannot be applied at very low T where the spin resistivity in experiments is dominated by effects other than s − d scattering model of the present paper. One of these possible effects is the carrier proliferation with increasing temperatures in semiconductors which makes the resistivity decrease with increasing T experimentally observed in magnetic semiconductors at low T .	2010-12-29T20:03:28.000Z	2009-12-05T00:00:00.000	{ "year": 2009, "sha1": "635ed6bd5b0ac5bd4011f9744ae0de4bf6d42841", "oa_license": null, "oa_url": "http://arxiv.org/pdf/0912.0989", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "253b7fbaa099f5f6e5de55faa0ecfcad343b88c6", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
4107777	pes2o/s2orc	v3-fos-license	Polyphenols in Colorectal Cancer: Current State of Knowledge including Clinical Trials and Molecular Mechanism of Action Polyphenols have been reported to have wide spectrum of biological activities including major impact on initiation, promotion, and progression of cancer by modulating different signalling pathways. Colorectal cancer is the second most major cause of mortality and morbidity among females and the third among males. The objective of this review is to describe the activity of a variety of polyphenols in colorectal cancer in clinical trials, preclinical studies, and primary research. The molecular mechanisms of major polyphenols related to their beneficial effects on colorectal cancer are also addressed. Synthetic modifications and other future directions towards exploiting of natural polyphenols against colorectal cancer are discussed in the last section. Introduction Epidemiological studies exhibiting protective effect of diets rich in fruits and vegetables against different types of cancer have drawn increased attention to the possibility of exploiting biologically active secondary metabolites of plants to fight against cancer. Among the vast array of phytochemicals, compounds called "polyphenols" constitute one of the most numerous and widely distributed groups, covering more than 10,000 different chemical structures [1]. Polyphenols (PP) are reported to have antioxidant, anticarcinogenic, antiatherosclerotic, anti-inflammatory, spasmolytic, hepatoprotective, antiviral, antiallergic, antidiarrheal, antimicrobial, and oestrogenic activity [2]. Colorectal cancer (CRC) is the third most common diagnosed cancer in men after lung and prostate cancer throughout the world. While in women CRC occupies the second position after breast cancer worldwide. Prevalence of CRC is 18% higher in developed countries than developing and undeveloped nations. People of more than 50 years old are more prone to be affected by CRC, and incidence in males is greater than in females. Although diet and Western lifestyle are still considered as being the main factors responsible for CRC, no specific food or other environmental agent has been identified as an exact causative factor [3]. Thus far, clearly identified types or causes of CRC are hereditary nonpolyposis colorectal cancer, familial adenomatous polyposis, inflammatory bowel diseases, human papillomavirus, and acquired immunodeficiency syndrome [4]. Although surgical resection remains the only curative treatment for CRC, an alternative approach to reduce the mortality rate is chemoprevention, use of synthetic or natural compounds in pharmacologic doses [5]. Colon cancers result from a series of pathologic changes that transform normal colonic epithelium into invasive carcinoma. Dietary PP affect these different cellular processes by acting as chemopreventive blockers. So far, only one review article that has been published concentrated on the effect of polyphenols on colorectal cell lines [6], and only a limited number of polyphenols have been considered. This review focuses on the updated research on a wider variety of polyphenols as applied to colorectal cancer. PP has been given in Figure 1 on the basis of the chemical structures of the aglycone portions and Figure 2 gives the basic structures of major groups [7]. Chemistry of PP and Their Dietary Sources A list of the 100 richest dietary sources of PP has been produced using comprehensive Phenol-Explorer data [8]. The richest sources are various spices and dried herbs, cocoa products, some dark coloured berries, some seeds (flaxseed) and nuts (chestnut, hazelnut), and some vegetables, including olive and globe artichoke heads. Top ten of the list containing the highest amount of PP is in the following order: cloves > peppermint (dried) > star anise > cocoa powder > Mexican oregano (dried) > celery seed > black chokeberry > dark chocolate > flaxseed meal > black elderberry. Pathogenesis of CRC and Its Signalling Pathways Acquired functional capabilities of cancer cells that would allow them to survive, proliferate, and disseminate are known as the hallmarks of cancer, that is, sustaining proliferative signalling, evading growth suppressors, resisting cell death, enabling replicative immortality, inducing angiogenesis, activating invasion and metastasis, reprogramming of energy metabolism, and evading immune destruction [9]. Underpinning these hallmarks are genomic instability and inflammation. While genomic instability confers random mutations including chromosomal rearrangements, causing genetic diversity that expedites the acquisition of hallmarks of cancer, the inflammatory state of premalignant and frankly malignant lesions that is driven by cells of the immune system also fosters multiple hallmark functions. Based on investigation of different stages of tumour initiation and progression, Fearon and Vogelstein proposed a model of colorectal carcinogenesis that correlated specific genetic events with evolving tissue morphology [10]. The Wnt/ -catenin pathway plays a dominant role in an initial stage of CRC development. Inactivation of the adenomatous polyposis coli gene is a key starting event in carcinogenesis of more than 60% of colorectal adenomas and carcinomas leading to stimulation of the Wnt pathway via free -catenin [10]. Stimulation of the epidermal growth factor receptor (EGFR) leads to the activation of KRAS or phosphatidylinositol-3-kinase pathways, which is important in CRC development from early adenoma to intermediate adenoma. Subsequently, numerous signal transduction molecules initiate a cascade of downstream effectors that trigger tumour growth, angiogenesis, and metastasis [11]. Transforming growth factor-(TGF-) is a multifunctional polypeptide that binds to specific TGF-receptors for paracrine and autocrine signalling. This ligand and receptor complex triggers intracellular signalling cascades that include the canonical Smad2 signalling pathway, which complexes with Smad4 and accumulates and translocates into the nucleus. In the nucleus, activated Smad complexes regulate the transcription of specific genes and ultimately regulate cell cycle and tissue repair [12]. TGF pathway contributes to a favourable microenvironment for tumour growth and metastasis throughout all the steps of carcinogenesis [13]. TGF-also induces apoptosis, from the association of death-associated protein 6 (DAXX) with the death receptor Fas. After binding, DAXX is then phosphorylated by BioMed Research International homeodomain-interacting protein kinase 2 (HIPK2), which then activates apoptosis signal-inducing kinase 1 (ASK1). ASK1 activates the Jun amino-terminal kinase (JNK) pathway that causes apoptosis [14][15][16]. Inactivation of TGF-beta pathway components is first detected in advanced adenomas and affects 40-50% of all CRCs [17]. Almost 50% of all CRCs show p53 gene mutations, with higher frequencies observed in distal colon and rectal tumours and lower frequencies in proximal tumours and those with the microsatellite instability or methylator phenotypes [18]. The mutations in p53 or the loss of its functionality occurs mainly at the transition from adenoma to cancer, and the frequency of alterations in the gene increases with the corresponding progression of the lesion [19]. CRC cells share many properties in common with stem cells which are conserved in both dormant and actively proliferating cancer cells [20]. On top of maintaining "stemness" characteristics, CRC cells with metastatic potential dissociate from the tumour mass and spread to other organs in the body [21]. This is achieved through a dedifferentiation program called epithelial-mesenchymal transition (EMT). This key developmental program allows stationary and polarized epithelial cells to undergo multiple biochemical changes that enable them to disrupt cell-cell adherence, lose apical-basal polarity, dramatically remodel the cytoskeleton, and acquire mesenchymal characteristics such as enhanced migratory capacity, invasiveness, and elevated resistance to apoptosis [22]. Adhesion molecules that maintain cell-cell contact in the differentiated tumour cells, such as E-cadherin, are downregulated in the undifferentiated cells, while molecules that impart invasive and migratory behaviour would be upregulated. To accommodate both the "stemness" and mesenchymal properties of invasive CRC cells, it has been proposed that CRC cells with metastatic potential are like "migratory stem cells" [23]. The EMT process is initially driven by three core groups of transcriptional regulators described as follows. The first is a group of transcription factors (TFs) of the Snail zinc-finger family, including SNAI1 and SNAI2 (SLUG) [24]. The second group is the distantly related zinc-finger E-box-binding homeobox family of proteins ZEB1 and ZEB2 (SIP1) [25]. The third group is the basic helixloop-helix (bHLH) family of transcription factors, including TWIST1, TWIST2, and E12/E47 [26]. In CRC, 85% of resected specimens have moderate to strong TWIST1 expression [27]. The earlier steps of the metastatic cascade EMT program include local invasion, intravasation, survival while transiting through the circulation, and extravasation. EMT programs are dynamically regulated, and during the last step of the metastatic cascade, colonization, carcinoma cells are thought to switch back to an epithelial state through the reverse process, mesenchymal-epithelial transition (MET) [28]. The final stage of the invasion-metastasis cascade, colonization, is likely to require adaptation of propagated CRC cells to the microenvironment of a distant tissue [29]. Increased matrix metallopeptidases (MMPs) expression and their activation generally promote hallmarks of CRC progression including angiogenesis, invasion, and metastasis and correlate with shortened survival. MMPs comprise a large family of at least 25 zinc-dependent endopeptidases capable of degrading all components of the extracellular matrix (ECM) and are categorized primarily by their structural features as gelatinases, collagenases, membrane-type, stromelysins, and matrilysins [30]. Intercellular adhesion molecule-1 (ICAM-1) is a 90-kDa cell surface glycoprotein that is known to be a member of the immunoglobulin gene superfamily of adhesion molecules. ICAM-1 expression is closely associated with metastasis and may be a useful indicator of prognosis in patients with colorectal cancer [31]. It is evident from the above discussion that the pathogenesis of CRC is characterized by regulatory pathways that are complex involving several layers of communication, cascades, crosstalk, and extensive networking. CRC usually develops through interaction of cytokines, the chemical mediators of inflammation; cytokine receptors, present on the surface of a variety of cell types; secondary messengers which convey signals from cell surface to the interior; transcription factors, which regulate the expression of several genes that affect CRC. Figure 3 depicts the signalling pathways involved in CRC. Roles of PP in CRC Related to Chemoprevention and Apoptosis Consumption of PP rich food proved to be beneficial in occurrence of CRC in a national prospective cohort study [32]. Numerous studies have evaluated the efficacy of dietary polyphenols against CRC in vivo, in vitro model and in clinical trials [33][34][35]. Polyphenols can affect the overall process of carcinogenesis by several mechanisms and cause tumour cell death through apoptotic pathway. PP have been shown to be highly effective in scavenging singlet oxygen and various free radicals, which leads to DNA damage and tumour promotion [36]. PP also displayed chemopreventive effect through their impact on the bioactivation of carcinogens. Most carcinogens of chemical origin undergo biotransformation by Phase I metabolizing enzymes to be converted into more reactive form suitable for binding with DNA and proceed towards carcinogenesis process. PP were found to inhibit cytochrome P450 enzymes of the CYP1A family and thus act as chemopreventive agents [37]. On the other hand, by increasing the activity of Phase II metabolizing enzymes (glutathione reductase, glutathione peroxidase, glutathione S-reductase, catalase, and quinone reductase), PP are able to provide beneficial effects against CRC [38,39]. For example, PP obtained from apple inhibited growth of HT-29 human colon cancer cells by modulating expression of genes (GSTP1, GSSTT2, MGST2, CYCP4F3, CHST5, CHST6, and CHST7) involved in the biotransformation of xenobiotics [40]. Apoptosis is a vital physiological process in the normal development, and induction of apoptosis is highly anticipated mode as a therapeutic strategy for cancer control [44][45][46]. Bcl family of protein, caspase signalling proteins, and p53 genes are the key factors that regulate apoptosis [47]. PP are effective general inhibitor of cancer cell growth and inducers of apoptosis in different cancer cell lines, including leukaemia, skin, lung, stomach, colorectal, and prostate cancer cells [34,[48][49][50][51][52]. Anthocyanin, ellagic acid, curcumin, flavone induced apoptosis in various colon cancer cell lines by different mechanisms in miscellaneous observations [34,[53][54][55]. PP can prevent the DNA damage caused by free radicals or carcinogenic agents through diverse mechanisms: (a) direct radical scavenging [56,57], (b) chelating divalent cations involved in Fenton reaction [58], and (c) modulation of enzymes related to oxidative stress (glutathione peroxidase, glutathione reductase, superoxide dismutase, nitric oxide synthase, lipooxygenase, xanthine oxidase, etc.) [59]. Dietary PP can also act as prooxidants depending on the cell type, dose, and/or time of treatment, as they can enhance reactive oxygen species production and therefore induce apoptosis [58,60,61]. In colon cancer HT-29 cells, flavone enriched the mitochondrial pyruvate or lactate uptake, which augmented the superoxide radical production and led to apoptosis [62]. Recent Update of Key PP as Applied to CRC Reported antitumour activity of PP against CRC is largely based on in vitro studies, rodent model studies, and even human clinical trials. During in vitro studies on antitumour activity of PP, different colorectal cancer cells (HT-29, SW480, Caco-2, Colo-205, Colo-115, HCT-115, HCT-116, DLD-1, LoVo etc.) were cultured, and cell viability was determined via MTT [3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide] reduction assay [154], SRB (Sulforhodamine B) colorimetric assay [155], and crystal violet method [156]. In vivo animal model models were produced by inducing tumour chemically, resulting in APC Min/+ mouse and rodent xenograft models. Carcinogen [azoxymethane (AOM), dimethylhydrazine (DMH), dextran sodium sulphate (DSS)] induced colon cancer in rodents can recapitulate in a highly reliable way and frequently used to assess activity of PP. Mutations in the adenomatous polyposis coli (APC) gene are required to initiate familial adenomatous polyposis (FAP) and are also important in CRC tumorigenesis. Several studies have been conducted with PP in APC Min/+ mouse that contains (multiple intestinal neoplasia Min) a point mutation in the APC gene and develops numerous adenomas. The role of PP has also been investigated in xenograft model where human tumours are injected and established in immunodeficient mouse strains (nude or SCID mice). This section contains the outcome from the studies conducted with PP against CRC. Benzoic Acid Derivatives. From literature, only gallic acid among benzoic acid derivatives showed anticancer activity against CRC in vitro and in vivo model [157,158], but no study has been conducted to identify the anticancer mechanism of gallic acid in CRC. However, gallic acid is believed to exhibit its anticancer effect by upregulating Bax and downregulating Bcl-2 in other tumour models [157]. Vanillic acid showed significant activity with IC 50 values less than 30 M in three different CRC cell lines but mechanism has not been studied [159] although vanillic acid and protocatechuic acid did not show significant anticancer activity against CRC [160,161]. Cinnamic Acid Derivatives. Caffeic acid showed apoptotic cell death against HCT 15 cell lines although IC 50 value was very high (800 M). Similar findings were made by other researchers [162,163]. In a recent study caffeic acid did not show any significant activity against HT-29 cell lines up to 200 M concentration nor did chlorogenic acid [164] that did not show any significant activity against different human colorectal carcinomas [161]. IC 50 values of p-coumaric acid against some other CRC cell lines were around 1 mM and apoptosis was the mechanism of cell death [163,165,166]. Ferulic acid inhibited CRC progression at adhesion and migration steps but no IC 50 value was greater than 1 mM concentration [163]. Carnosic acid showed IC 50 values in the range of 24-96 M against Caco-2, HT29, and LoVo cell lines. It inhibited cell adhesion and migration, possibly by reducing the activity of secreted proteases such as urokinase plasminogen activator and metalloproteinases. These effects may be mediated through a mechanism involving the inhibition of the COX-2 pathway [167]. Sinapic acid showed IC 50 values of less than 25 M in three different CRC cell lines but mechanism has not been studied [159]. Isoflavones, Neoflavonoids, and Chalcones. Among isoflavones, biochanin A showed ID 50 values below 15 g/mL against two CRC cell lines and was found to enhance radiotoxicity in vitro [170,171]. Formononetin that showed dose dependent cell killing, both in vitro and in vivo in RKO cell line, induces apoptosis by modulating Bax/Bcl-2 activities, inactivating ERK pathway and TNF-/NF-B pathway [172]. Formononetin also showed anticarcinogenic activity in HCT-116 cells via promotion of caspase-dependent apoptosis and inhibition of cell growth, with contribution by downregulation of the antiapoptotic proteins Bcl-2 and Bcl-xL [173]. Daidzein killed 50% of HCT cells, LoVo cells, and DLD-1 cells at concentration of 40 M, 68.8 M, and 46.3 M, respectively, but against LoVo cells it exhibited biphasic effects by killing cells in dose dependent manner at higher concentrations (≥5 M) and vice versa at lower concentrations (≤1 M) [75,174,175]. Most commonly studied isoflavone, genistein, showed cytotoxicity against HCT, LoVo, and DLD-1 cell lines with IC 50 values of 15 M, 57.3 M, and 56.1 M, respectively, whereas in HCC-44B2 cells and HCC50-D3 the value was 11.5 g/mL and 9.5 g/mL [75,170,174]. Genistein reduces the density of cell surface charge and increases the order in membrane protein conformation which might be one of the mechanisms of its anticancer effect [174]. No literature reporting neoflavonoids activity against CRC was found. Among chalcones, phloretin caused apoptotic cell death to HT-29 cells with an IC 50 value close to 100 M. The mechanism involved changes in mitochondrial membrane permeability and activation of the caspase pathways [176]. Phloretin also has the potential to increase adoptive cellular immunotherapy against SW-1116 CRC cells [177]. Xanthohumol, another important chalcone, is found to show cytotoxicity in different CRC cells in vivo and in vitro with IC 50 values less than 5 M [178][179][180]. The apoptosis involved downregulation of Bcl-2, activation of the caspase cascade, and inhibition of topo I activity. In combination with chemotherapy, it is recommended for use in HCT-15 cell lines, being aimed to reduce drug resistance by inhibition of efflux transporters [180]. Xanthohumol inhibits metastasis by inhibiting expression of CXCR4 chemokine receptor [181]. Flavones, Flavonols, Flavanones, and Flavanonols. Among all different types of flavones, apigenin and luteolin were most commonly investigated phytochemicals for their anticancer activity against CRC. Important flavones that have been studied against CRC are given in Table 1. Among all different types of flavonols quercetin, chrysin and rutin were studied most for their anticancer activity against colorectal cancer models. Important flavonols that have been investigated against CRC are given in Table 2. Naringenin appears to be the most commonly studied phytochemicals among the flavanones that can act against colorectal cancer. It suppressed colon carcinogenesis through the aberrant crypt stage in azoxymethane-treated rats [74]. Another study showed that antiproliferative activity of naringenin was estrogen receptor dependent [182], while other in vitro studies gave mixed results in different CRC cell lines [65,80,152,183]. Another flavanone, hesperetin, significantly reduced the formation of preneoplastic lesions and effectively [70] Apigenin HCT116 Induced cell death due to apoptosis and autophagy where apoptosis is via decreased expression of cyclin B1, Cdc2, and Cdc25c; increased expression of p53 and p21CIP1/WAF1; decreased levels of procaspase-8, -9, and -3. [71] HT-29 and HCT-15 Oxidative stress resulted in senescence and chemotherapeutic effect. [72] SW480 and HCT-15 Suppressed cell proliferation, migration, and invasion via inhibition of the Wnt/ -catenin signalling pathway. [ Downregulated the activation of the PI3K/Akt and ERK1/2 pathways via reduction in IGF-IR signalling which may be one of the mechanisms responsible for the observed apoptosis and cell cycle arrest. [78] Luteolin HT-29, SW480 In HT-29 cells, IC 50 value was greater than 200 M but in SW480 cells it is 90 M. [79,80] Male Balb/c mice Inhibited azoxymethane-induced colorectal cancer growth through activation of Nrf2 signalling; altered carbohydrate metabolizing enzymes; decreased expressions of iNOS and COX-2; restored reduced glutathione and protein thiols; decreased lysosomal enzymes, induced apoptosis by modulating Bcl2, Bax, and caspase-3; decreased mucin depleted foci, levels of glycoconjugates; controlled cell proliferation by inhibiting wnt/ -catenin/GSK-3 pathway. Luteolin also acts as antimetastatic agent by decreasing MMP-9 and MMP-2. [100] HT-29 Resveratrol and quercetin in combination showed anticancer activity in colon cancer cells and repressed oncogenic microRNA-27a. [108] HT-29 xenografts in female nude mice Quercetin and trans-pterostilbene in combination facilitated elimination of colorectal cancer by chemoradiotherapy through a Bcl-2-and superoxide dismutase 2-dependent mechanism. [109] CF1 mice, F344 rats, Wistar rats Azoxymethane and dimethylhydrazine induced colon cancer study showed reduction of aberrant crypt foci and focal areas of dysplasia. [113] HCT-116 Chrysin sensitized tumour necrosis factor--induced apoptosis in human tumor cells via suppression of nuclear factor-kappaB. [122] HCT-116 Promoted tumour necrosis factor-(TNF-) related apoptosis-inducing ligand (TRAIL) induced apoptosis. [123] SW480 Chrysin caused cell-cycle arrest at the G2/M phase in a dose-dependent manner. [80] HCT116, DLD1 and SW837 Aryl hydrocarbon receptor was required for the chrysin induced apoptosis and the upregulation of TNF-andgene expression and consequent activation of the TNF-mediated transcriptional pathway. [124] Caco-2 Blocked topotecan-induced apoptosis in spite of inhibition of ABC-transporters. [125] Kaempferol SW480 Sensitized TRAIL-induced apoptosis. [126] HCT-116 The IC 50 of kaempferol was 53.6 M in HCT116 (p53+/+) cells and 112.7 M in HCT116 (p53−/−) cells. It induced via ataxia-telangiectasia mutated-p53 pathway with the involvement of p53 up-regulated modulator of apoptosis. [127] HT-29 Kaempferol increased chromatin condensation, DNA fragmentation, and the number of early apoptotic cells in a dose-dependent manner. Kaempferol increased the levels of cleaved caspase-9, caspase-3, and caspase-7 as well as those of cleaved poly (ADP-ribose) polymerase. Moreover, it increased mitochondrial membrane permeability and cytosolic cytochrome c concentrations. [128] Isorhamnetin HT-29, FVB/N mice Chemoprotective effects of isorhamnetin were linked to its inhibition of oncogenic Src activity and consequential loss of nuclear -catenin, activities that were dependent on CSK expression. Had IC 50 value less than 350 g/mL after 48 h and induced apoptosis by modulation of Bcl-2 family members and Fas receptor. [136] modulated the xenobiotic-metabolizing enzymes in rats during DMH-induced colon cancer study [184,185]. Among flavanonols, only taxifolin acts as an effective chemopreventive agent against colon carcinogenesis due to its antioxidant mediated apoptosis and antiproliferative activities [186,187]. Taxifolin is found to control NF-kB-mediated Wnt/ -catenin signalling via upregulating Nrf2 pathway [188]. HCT116, HT29, SW480, and SW837 cell lines, respectively. Mechanism of action has been linked to the inhibition of growth and activation of the epidermal growth factor receptor and human epidermal growth factor receptor-2 signalling pathways [189]. Among eleven different types of flavanol, EGCG showed the highest antiproliferative activity against HCT-116 cells at 50 M [190]. In APC Min/+ mice, EGCG significantly inhibited intestinal tumorigenesis. Oral administration of EGCG showed increased levels of E-cadherin and decreased levels of nuclear -catenin, c-Myc, phospho-AKT, and phosphoextracellular signal-regulated kinase 1/2 (ERK1/2) in small intestinal tumours [191]. In another mice model, EGCG reduced inflammation-related colon carcinogenesis induced by azoxymethane and dextran sodium sulphate. Antitumour activity has been ascribed to decrease in mRNA expression levels of COX-2 and inflammatory cytokines (TNF-, IFN-, IL-6, IL-12, and IL-18) in the colonic mucosa [192]. Other studies mentioned the proposed anticancer mechanisms of EGCG including cell cycle arrest and apoptosis through inhibition of cyclooxygenase-2 expression, activation of AMPactivated protein kinase, cyclin D1 degradation and p21 transcriptional activation, and inhibition of HES1 and Notch2 signalling in different colorectal cancer cell lines [43,[193][194][195][196]. EGCG also inhibited invasion and MMP expression, angiogenesis, through blocking the induction of VEGF [197,198]. Molecular mechanism for antitumour activity of EGCG and quercetin is shown in Figure 4. Flavanols and Chemopreventive effects of theaflavin have been reported in azoxymethane induced colon cancer study in male Sprague Dawley rats [199]. The mechanism behind the beneficial effects of proanthocyanidins against CRC has been related to inhibition of angiogenesis through suppressing the expression of VEGF and Ang1 [200], induction of apoptosis [201], and antioxidant activity [202]. Polyphenolic Amides. Capsaicin is the most studied polyphenolic amide against CRC. In vitro and in vivo studies in mice bearing Colo-205 tumour xenografts showed that capsaicin significantly reduced tumour progression by activating caspase-3, caspase-8, caspase-9, Bax, Fas and reducing Bcl-2 [218]. Capsaicin and 3,3 -Diindolylmethane worked synergistically against CRC via modulating transcriptional activity of NF-B, p53, and target genes linked with apoptosis [219]. Other studies showed that anticancer action of capsaicin was related to nitric oxide production, reactive oxygen species generation, and suppression of transcriptional activity of -catenin/TCF pathway [220][221][222]. In our study, we have found synergism in binary sequenced combination of capsaicin with oxaliplatin in all sequences (0,0 h; 0,4 h; 4,0 h) and more so in higher concentration in Lim 2405 cell line (unpublished data). Dihydrocapsaicin, a saturated structural analogue of capsaicin, was found to possess greater activity than capsaicin against HCT-116 cells and induced autophagy in a catalase-regulated manner [223]. Avenanthramides significantly inhibited proliferation of HT29, Caco-2, LS174T, and HCT116 human colon cancer cells [224]. Other Polyphenols 5.4.1. Resveratrol. 3,5,4-Trihydroxystilbene, known as resveratrol, is one of the most studied polyphenols against CRC. It entered into clinical trial after a number of preclinical studies for its encouraging activity and nontoxicity. All studies conducted in rodent model and clinical trial regarding the activity of resveratrol up to 2009 have been described by Bishayee in his review [225]. The authors mentioned 9 in vivo studies related to CRC, among which three were conducted in APC Min/+ mice model and others related to chemically induced tumour [226][227][228][229][230][231][232][233][234]. All the studies discussed in the review (except one on APC Min/+ mice study) provided support for therapeutic potential of resveratrol against CRC. Later on, in 2012, Juan et al. published another review article that focused on the effects of resveratrol on CRC from conducted in vivo studies and clinical trials [235]. The authors provided details on the molecular mechanism of action of resveratrol against CRC. Two more research articles recently described the promising effect of resveratrol in mouse model against CRC, which were not mentioned in earlier reviews [236,237]. The results from the investigations of the anticancer activity of resveratrol against CRC have not been detailed here because they have been considered well in the previously mentioned reviews. However, a pictorial representation of the molecular mechanism of action of resveratrol against CRC is given in Figure 5. The effect of 5-fluorouracil increased in combination with resveratrol due to chemosensitizing property [238]. Antimetastatic activity of resveratrol in CRC has also been reported [237,239]. Since resveratrol is found to downregulate multidrug resistant protein 1 by preventing activation of NF-B signalling and suppressing cAMP-responsive element transcriptional activity, it can be used to overcome drug resistance by combining with other chemotherapeutic drugs [240]. We have found synergism at higher concentration in binary sequenced combination (at 0,0 h; 0,4 h; 4,0 h) of resveratrol with cisplatin but additive to antagonism at lower concentration in HT-29, Caco-2, and Lim 2405 cell lines (unpublished data). Curcumin. Curcumin is the main active compound in turmeric (dried rhizome of Curcuma longa). We have found 82 research articles (in vitro and in vivo preclinical studies, clinical trial) describing effect of curcumin including mechanism of action against CRC. Here, we have not considered the anticancer activity of curcumin in detail because three review articles discussed well the therapeutic potential of curcumin for CRC along with its mechanism of action [241][242][243]. Curcumin can inhibit the initiation of carcinogenesis by increasing glutathione S-transferase, induce cell cycle arrest in S and G2/M phase, induce apoptosis, and inhibit metastasis by decreasing CD31, VEGF, IL-8, and mir-21 expression [244][245][246][247]. Mechanism of curcumin in CRC as applied to apoptosis is shown in Figure 6. Curcumin has proved to be beneficial in combination with chemotherapy and radiotherapy as well [248,249]. Synergistic activity of curcumin has also been observed in our study in different sequences and doses with oxaliplatin and cisplatin in four colorectal cell lines (unpublished data). Rosmarinic Acid and Gingerol. Rosmarinic acid is the main component of Rosemary which at high dose showed antitumour activity applying to in vitro and DMH-induced in vivo study against CRC [250,251]. It is thought that MAPK/ERK pathway is linked with the apoptotic mechanism of rosmarinic acid in CRC [252]. Rosmarinic acid has also been reported to possess antimetastatic activity [253]. 6-Gingerol is the most important ingredient of ginger showing antiproliferative activity in a dose dependent manner against LoVo cell lines via G2/M cell cycle arrest. Exposure to 6-gingerol induced intracellular ROS and upregulate p53, p27Kip1, and p21Cip1 levels leading to consequent decrease of CDK1, cyclin A, and cyclin B1 [254]. In HCT-115 cell line, its anticancer action was found to be mediated by inhibition of Leukotriene A4 hydrolase [255]. Another study showed that 6-gingerol stimulated apoptosis through upregulation of NAG-1 and G1 cell cycle arrest through downregulation of cyclin D1 that involved protein degradation as well as -catenin, PKC , and GSK-3 pathways [256]. However, 6-gingerol did not show anticancer activity in Colo-205 cell line [257]. Ellagic Acid, Secoisolariciresinol, and Matairesinol. Ellagic acid is a dilactone of hexahydroxydiphenic acid that occurs naturally in berries and nuts such as the raspberry, strawberry, walnut, and pecan. It inhibited growth of Caco-2 cell line, possibly mediated by regulation of matrix metalloproteinases, vascular endothelial growth factor expression, and induction of apoptosis [258]. According to others, the anticancer action is mediated through downregulation of cyclins A and B1 and upregulation of cyclin E, cell cycle arrest in S phase, induction of apoptosis via intrinsic pathway (Fas-independent, caspase 8-independent) via Bcl-XL downregulation with mitochondrial release of cytochrome c into the cytosol, and activation of initiator caspase-9 and effector caspase-3 [54]. Studies on DMH-induced colon carcinogenesis also proved the beneficial effect of ellagic acid and showed the mechanism to be linked with reduced expressions of NF-, COX-2, iNOS, TNF-a, and IL-6 as well as inhibition of AKT-phosphoinositide-3 kinase pathway [259,260]. However, metabolic products urolithins were found to be more potent against CRC compared to ellagic acid itself [261]. Mixed urolithins and ellagic acid inhibited phenotypic and molecular colon cancer stem cell features as well [262]. Secoisolariciresinol and matairesinol are lignans. They constitute one of the major groups of phytoestrogens that have been investigated against CRC but the activity remains controversial. Earlier study with flax seed diet containing secoisolariciresinol and matairesinol showed significant antitumour activity [263]. However, in vivo and in vitro studies done later did not provide insights into the anticancer potential of secoisolariciresinol and matairesinol [264,265]. Chemical Modifications of PP in Nature and Synthetic Analogues The most important impediment to the successful development of natural PP as clinical therapy against CRC is their low bioavailability. To overcome the problem towards reaching therapeutic concentrations of PP and increasing efficacy, many researchers tried to produce a number of synthetic analogues through structural modifications. Increase in potency in vitro and bioavailability in vivo has been observed in many studies. Some selected reports concerning chemical modifications of PP applied against CRC are represented in Table 3. Combination of Polyphenol with Chemotherapy/Radiotherapy and Other PP Emerging evidence suggests that a single-agent approach is probably less likely to be very effective in the management cancer. The rationale for recommending a multidrug regimen is to attack cancer cells through multiple targets and diverse mechanisms of actions with reduced toxicity, ultimately leading towards improved clinical outcomes. With that aim, PP have been investigated with chemotherapy and radiotherapy in various cancer models. For example, curcumin given in combination potentiated the cytotoxic effects of doxorubicin, 5-FU, and paclitaxel in prostate cancer cells [266]; enhanced the antitumour activities of cisplatin, doxorubicin, and Taxol in HA22T/VGH hepatic cancer cells, HeLa cells, or CAOV3 and SKOV-3 ovarian cancer cells [267,268]; sensitized multiple myeloma cells to vincristine and melphalan [269]. Similar evidence is available in literature for resveratrol, EGCG, quercetin, genistein, proanthocyanidin, and daidzein in various types of cancer with different classes of chemotherapeutic drugs [169,[270][271][272][273]. A few studies also have been conducted against CRC where it has been observed that PP in combination with other chemotherapeutic drugs produced synergism, for example, curcumin with 5-fluorouracil against HCT-116 and HT-29 cell line, resveratrol metabolites and oxaliplatin in SW-480 and SW-620 cell lines, genistein with 5-fluorouracil in HT-29 cell line [274][275][276][277]. Outcomes of some other combination studies already have been mentioned in describing the effect of individual PP. Like chemosensitization, PP also have shown the potential of radiosensitization in various cancers, but investigations of the same effects against CRC is very scarce [278][279][280][281]. In one study quercetin has been shown to increase chemoradiosensitivity against colorectal cancer in xenograft mouse model [109]. Plenty of evidence exists in literature regarding the benefits of the effect of one polyphenol combined with another polyphenol against different types of cancer. Combination of resveratrol with black tea polyphenols resulted in a synergistic tumour suppressive response in mouse skin tumour [282]. Resveratrol showed better chemopreventive response when combined with curcumin by maintaining adequate zinc and modulating Cox-2 and p21 level in mouse model of lung cancer [283]. Combination of genistein with resveratrol reduced the most severe grade of prostate cancer in SV40 Tag-targeted probasin promoter rat model [284]. EGCG in combination with curcumin synergistically inhibited oral premalignant epithelial cells [285]. Quercetin and resveratrol in combination with ellagic acid showed synergism against leukaemia [286]. However, studies on the effect of combination of pure polyphenols against CRC are not numerous. Curcumin showed synergistic antitumour effects in combination with resveratrol in one report of colon cancer model in SCMID mice [287]. Combination of epicatechin and EGCG also exhibited synergistic outcome against HT-29 cancer cells. There are few reports in literature related to the beneficial effects of plant extracts or juices that possess mixture of polyphenols against CRC [288,289]. Bioprospecting and molecular pharmacology studies have shown that PP can modulate the survival pathways induced by cancer cells, carcinogens, and chemotherapeutics. The possible mechanisms of chemoresistance are shown in Figure 7. In Section 5, we have considered the molecular mechanisms by which PP would produce anticancer action in CRC. It is thought that PP have the ability to effectively modulate the various mechanisms of chemoresistance. For example, EGCG and quercetin can directly inhibit PI3/AKT pathway, NF pathway, EGFR family pathway, and IAP family pathway and increase p53 (Figure 4). Similarly curcumin can inhibit Bcl-2 family pathway, EGFR family pathway, and NF pathway ( Figure 6). In terms of IC 50 values, synthetic analogue found to be more sensitive against 3 CRC cell lines. In vivo study also proved its greater activity. [137] Resveratrol 3, 5, 4 -Trimethoxystilbene, 3, 3 , 4, 5 -tetramethoxystilbene Inhibited HT-29 cell growth. [138] Digalloyl resveratrol Inhibited HT-29 cell growth more effectively than gallic acid and resveratrol. [139] Flavone 3 , 4 , 5 , 5, 7-Pentamethoxyflavone More active compared to tricin and apigenin in APC Min/+ mice model. [140] Curcumin Dimethoxycurcumin In HCT-116 cell lines, dimethoxycurcumin is more potent in terms of ability to kill cancer cells by apoptosis, less extensively metabolized in microsomal systems, and more stable in vivo compared to curcumin. [141] Curcumin difluorinated At higher concentration synthetic analogue showed greater potency than curcumin in HCT-116 cells. [142] EF31 and UBS109 Both analogues showed significant antitumour activity in colorectal xenograft model possibly via inhibition of NF-B and cell cycle progression at G0/G1 phase. [143] GO-Y030, FLLL-11, and FLLL-12 All of the analogues exhibited 4 to 20 times greater activity than curcumin against SW480, HT-29, and HCT116 cell lines but with minimal toxicity against normal cell line. [144] EGCG Peracetylated EGCG Administration of the synthetic analogue was more effective than EGCG in preventing the shortening of colon length and the formation of aberrant crypt foci and lymphoid nodules in mouse. [145] Procyanidin dimer [150] Current Status of PP in Clinical Trials Related to CRC Following the discovery of significant anticancer potential of curcumin, resveratrol, EGCG, and genistein seen in studies related to in vitro and in vivo rodent model, the compounds entered into clinical trials for efficacy and toxicity study in human model. Many of the reported Phase I, Phase II, and Phase III studies further validated the potential of using PP against CRC. Benamouzig and Uzzan provided a summary of 21 clinical trials conducted by 2016 related to the use of curcumin against CRC in their review [243]. Similarly 17 clinical trials on resveratrol have been reviewed elsewhere [290]. Few other chemotherapeutics and chemoprevention clinical trials conducted on PP have been summarized by Vinod et al. [169]. Table 4 represents the recently completed or ongoing clinical trials on PP against CRC, which have not been covered by others. Future Perspective and Directions From our study we have found that curcumin, resveratrol, quercetin, luteolin, apigenin, EGCG are the most investigated polyphenols against CRC. In terms of in vitro cytotoxicity these polyphenols gave average IC 50 value around 15-60 M against different colorectal cancer cell lines, which is comparatively larger than clinically used anticancer drugs. Moreover some results from single administration of curcumin or resveratrol produced contradictory evidence against in vitro and in vivo model data. It would be unwise to be overoptimistic and battle against CRC with a single polyphenol only. Rather the strength of polyphenols can be exploited by combining them with clinically used chemotherapeutic drugs to reduce dose related side effects of chemotherapy and overcoming drug resistance. Therefore, we can search among polyphenols for the ones that give synergistic effect with chemotherapeutic drugs or with other phytochemicals. In our laboratory we have been working with curcumin, resveratrol, EGCG, quercetin, capsaicin, 6-gingerol, genistein in combination with cisplatin and oxaliplatin against different CRC cell lines. In many cases we found synergism (unpublished data) like others [291]. Chemical modifications of natural PP can be continued to improve their activity and bioavailability. Another area of research that could be explored is related to exploiting ability of PP to interact with stem cells in CRC. Cancer stem cells are multipotent cells that possess self-renewal capacity and high proliferative capacity and lead to metastasis through migration. Although cancer stem cells represent less than 2.5% of the tumour mass, they may be responsible for the resistance to cancer therapies and relapse in CRC [292]. Wnt/ -catenin, Hedgehog, and Notch have been identified to play pivotal roles in cancer stem cells self-renewal. Presently researchers are targeting the hallmark stem cell-like properties of tumour cells to overcome cancer. A number of phytochemicals have also been investigated against cancer stem cells in several studies. Curcumin suppressed mammosphere formation along serial passage in breast cancer, and the effect of curcumin on breast cancer stem/progenitor cells was seen to be mediated through its potent inhibitory effect on Wnt/ -catenin signalling [293]. Genistein also reduced breast cancer stem cells by inhibiting AKT and increasing PTEN [294]. Likewise, resveratrol inhibited pancreatic cancer stem cell characteristics in human and mouse model by inhibiting EMT [295]. In pancreatic cancer model quercetin decreased ALDH1 activity, induced apoptosis, and reduced the expression of proteins implicated in EMT in vitro, while it inhibited stem cell-derived xenografts in vivo, reducing the expression of proliferation, stemness, and angiogenesis related genes [296]. However, very little has been studied to modulate the stem cells by PP in CRC. Conclusion As oxidative stress is an inescapable part of aerobic life, it can be said that cancer with its origin in mutations is a disease of living even though it evokes death sentence in many minds. However, as nature creates problems, it also provides solutions. As tumour active polyphenols have been a part of human diet for thousands of years, but without any adverse side effect, it is thought that selected tumour active polyphenols or their derivatives in combination with targeted therapy may provide an affordable means of overcoming drug resistance and reducing side effects in colorectal cancer and indeed in many other cancers. Conflicts of Interest Md Nur Alam, Muhammad Almoyad, and Fazlul Huq declare that they have no conflicts of interest. Authors' Contributions Muhammad Almoyad conducted the literature review, Md Nur Alam has drafted the entire review, and Fazlul Huq supervised the whole process and critically reviewed the manuscript.	2018-04-03T04:09:58.496Z	2018-01-15T00:00:00.000	{ "year": 2018, "sha1": "48c67edb4e8c5ffa92a0b64aef0070c371e8e3f2", "oa_license": "CCBY", "oa_url": "http://downloads.hindawi.com/journals/bmri/2018/4154185.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "dd2d69050bd94e131c849ac2857059722eed27a4", "s2fieldsofstudy": [ "Medicine", "Chemistry" ], "extfieldsofstudy": [ "Medicine" ] }
156674280	pes2o/s2orc	v3-fos-license	A COMPARISON OF SAME SLOPE SEASONALITY AND HOLT- WINTERS SMOOTHING FORECASTING MODELS In this paper analytical expression for determining the optimal parameter value of the Same Slope Seasonality model was developed. Then performance of the Same Slope Seasonality model was compared to the performance of Holt-Winters exponential model, performing tests on M2-Competition time series. To determine the parameters of Holt-Winters exponential model, nonlinear mathematical programming was used. Performed tests proved that Same Slope Seasonality model is more successful than Holt-Winters exponential model. Tests revealed that Same Slope Seasonality model gives unreliable forecasts when time series has specific charac teristics, giving us valuable information for model improvement. Introduction Forecasting model reliability is not conditioned by its complexity, and simple forecasting models can be accurate or even more accurate than more complex models [1]. According to [2] increasing complexity leads to reduced accuracy regardless of forecasting method used. In the works which were the subject of the study in [2], and where the increase in errors could be quantified, authors report that an increase in complexity increases the prediction error for about 25%. Authors recommend that forecasters should use those forecasting models that they can understand. Exponential smoothing models because of their characteristics have found very wide application in production and service organizations [1,3]. Some authors because of their characteristics such as simplicity, ease of use and very good reliability, highlight these families of forecasting models even in relation to more complex such as ARIMA models [4,5]. SS This paper deals with Same Slope (SS) and Same Slope Seasonality (SSS) forecasting models. SS and SSS forecasting models are of recent. In comparation to referent exponential smoothing models, SS and SSS are simpler models. Also, SS and SSS forecasting models are according to conducted studies [6,7,8] give forecasts of better reliability compared to exponential smoothing models. Aim of this paper is to try to make SS and SSS forecasting models even easier to implement. In this paper, analytical expression for determining the optimal value of forecasting model parameter was developed. This work is an attempt to make SSS model even easier in terms of implementation. The existence of analytical expression for determining the optimal parameter value now does not require any minimization of the standard deviation of the forecasting model. Determining optimal forecasting model parameter value has so far required defining and solving of the nonlinear mathematical programming model. In this study we also compared performance of the SSS model to the performance of Holt-Winters (HW) exponential forecasting model. To compare performances HW forecasting model is chosen as very popular model among practitioners and researchers [9]. Tests were performed on the time series of M2-Competition [10]. Settings of SS forecasting model are presented in [7]. In the same study, tests were carried out on the real time series. SS model is a single parameter model, simple to understand and apply. In [8] tests and performance comparation of the SS model to fifteen models of exponential smoothing have been performed. The results showed that the developed model is more efficient than models of the same level of complexity. Main settings and results of performed tests of SSS forecasting model are presented in [6]. SSS forecasting model is still the single parameter model based on the same logic as SS model. The results showed that the developed model generates forecasts of the same level of reliability as well as more complex exponential smoothing models. Settings of the same slope and same slope seasonality models SS model is based on idea that time series will have the same gradient (slope) in the next time period like in previous. Model is defined by (1), whose settings are presented in [7]: Graphical presentation of logic of SS model is given on Fig. 1. (2), whose settings are presented in [6]: With respect to parameter  following cases can be considered: 1   -forecast smaller change rate in the period t to tm  than in the prior period, e.g. forecasts smaller change rate in the period t to tm  than in the period tp  to t p m . 1   -forecast the same change rate in the period t to tm  as in the period tp  to t p m . 1   -forecast bigger change rate in the period t to tm  than in the period tp  to t p m . Considering (1) and (2) it can be concluded that in the case 1 p  , SSS model becomes SS model. Because of this fact, only analytical expression for determining parameter  of SSS model is developed. Determining optimal value of the same slope seasonality model parameter SSS model because of its mathematical simplicity, allows finding expression for determining optimal  parameter. It is necessary to formulate function of squared deviations between forecasted and true values, dependent on  parameter is given by: Substituting (2) into (3) we get expression for sum of squared errors as the function of  parameter: Function   f  is convex and its minimum can be found by equaling its first derivative to zero. The first derivative of the function of squared errors is given by: Equaling the first derivative of   f  function to zero gives analytical expression for calculating  parameter: Where n is number of data in time-series. The value of  parameter calculated by (6) corresponds to the value determined using nonlinear mathematical programming, where standard deviation or MSE of model is minimized. Research methodology Performances of SSS model are compared to the performances of three-parameter HW exponential smoothing model with additive trend and additive seasonal component, which is listed in standard list of exponential smoothing models (1): Tests are performed on times series "M2CAll" from M2-Competition, available at [11]. Last year data have been used to estimate MAPE of the considered models, and all other data are used to estimate optimal values of forecasting models parameters. SSS model parameter is determined using (6) for case 1 m  . To determine HW model parameters, nonlinear mathematical programming is used. Nonlinear mathematical programming models are solved using Solver in Microsoft Excel application. The model standard deviation is used as the objective function of nonlinear mathematical programming model, also for the case 1 m  . Forecasts are generated for time horizon of one year. After forecasts are generated, true time series values are used to estimate reliability measures. To compare model performances MAPE is used as a measure of model reliability. Forecasts are generated using only quantitative data and without any available qualitative informations. Tests are performed on 28 of 29 time series from M2-Competition. Time series INTERSAL haven't been used in tests, because its value in the last year is 0, which prevents calculating MAPE value. Among 28 analyzed, 22 were monthly and 6 quarterly time series. Results The developed expression for determination of SSS model  parameter, for each of the tested time series, gives  value identical to the value obtained by using Solver in MS Excel, set to solve nonlinear programming model in a way that the problem is set up to search the  parameter which minimizes the standard deviation of the model. MAPE values for both models and all analyzed time series are given in Table 1 Table 1, that SSS forecasting model has shown better performances than more complex HW forecasting model on respectable number of time series. Important advantage of SSS in relation to HW forecasting model is existence of analytical expression for direct calculation of optimal  parameter value. For estimation of optimal parameter values of HW model it is necessary to use nonlinear mathematical programming and appropriate solver applications. Importance of developed model is even bigger if it is known that HW forecasting model is very robust and reliable, which in many studies shows close or better performances even than very complex ARIMA models [1,12,13,14]. While performing tests, we discovered very important disadvantage of SSS model, and it is that forecast results are very affected by last data in time series. This disadvantage is not present when time series has no non-standard observations. Model gives very poor results in case that last available data in time series is non-standard observation. As an example of unreliable forecast is forecast of PANTER time series, where SSS model had MAPE value of 259,65%. Poor forecast results do not mean model is unreliable, because for the same time series HW model gave even poorer results. It is obvious that random component is very dominant in this time series, and in order to do meaningful forecasts, some qualitative informations are required. Conclusion The developed expression for determination of SSS model  parameter, for each of the tested time series, gives  value identical to the value obtained by using Solver in MS Excel, set to solve nonlinear programming model in a way that the problem is set up to search the  parameter which minimizes the standard deviation of the model. SSS model is single parameter model having only one expression, while HW model is three-parameter model having three expressions, required for determination of model parameters. Comparative advantage of SSS model to HW model especially is important when considering costs of parameter determination. SSS model parameter is simply determined using single analytical expression, while HW model parameter determination requires solving nonlinear mathematical programming model. Besides this obvious advantage, SSS model is easier to understand and implement than HW model, and also it requires less computational operations and efforts in forecast generations. Forecasts generated by SSS model are of the same reliability levels as forecasts generated by more complex HW model. SSS model performed well in forecasting time series with dominant seasonal as well as trend component. Unreliable forecasts by SSS model can be expected in case of non-standard observations in the time series, because SSS model forecasts are very dependent on current observation for which forecasting value is generated. By introducing another parameter, which should relativize influence of single observation, maybe would improve disadvantage of SSS model that one last observation has such serious impact on forecast. Another suggestion is that instead of using value of one observation, value of moving average for some period length could be used.	2019-05-18T13:08:03.030Z	2016-02-15T00:00:00.000	{ "year": 2016, "sha1": "8ae3ecb853fc6dd8e44667ca6053794d7e617de5", "oa_license": null, "oa_url": "https://doi.org/10.2507/26th.daaam.proceedings.009", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "99df18721df9e91adda902e0ecb3a23040c0aff3", "s2fieldsofstudy": [ "Mathematics", "Economics" ], "extfieldsofstudy": [ "Mathematics" ] }
195246120	pes2o/s2orc	v3-fos-license	Automated monitoring of behaviour in zebrafish after invasive procedures Fish are used in a variety of experimental contexts often in high numbers. To maintain their welfare and ensure valid results during invasive procedures it is vital that we can detect subtle changes in behaviour that may allow us to intervene to provide pain-relief. Therefore, an automated method, the Fish Behaviour Index (FBI), was devised and used for testing the impact of laboratory procedures and efficacy of analgesic drugs in the model species, the zebrafish. Cameras with tracking software were used to visually track and quantify female zebrafish behaviour in real time after a number of laboratory procedures including fin clipping, PIT tagging, and nociceptor excitation via injection of acetic acid subcutaneously. The FBI was derived from activity and distance swum measured before and after these procedures compared with control and sham groups. Further, the efficacy of a range of drugs with analgesic properties to identify efficacy of these agents was explored. Lidocaine (5 mg/L), flunixin (8 mg/L) and morphine (48 mg/L) prevented the associated reduction in activity and distance swum after fin clipping. From an ethical perspective, the FBI represents a significant refinement in the use of zebrafish and could be adopted across a wide range of biological disciplines. Illustration of the Fish Behaviour Index (FBI) system The FBI takes data from the previous 1 minute interval and refreshes every 1 minute. In our system we currently have analysis time periods of 1, 10 and 30 minutes, which means that on a rolling 1 minute basis the data from the last 1 minute is reported at time period 1 minute, data from the last 10 minutes in the 10 minute period, and data from the previous 30 minutes in the 30 minute period. The data is therefore analysed in the following periods: 1, 2, 3, …, 30 minutes; 1-10, 2-11, 3-12, … 21-30 minutes; 1-30, 2-31, 3-32, …, t to t+29 minutes. Our videos were generally stopped at 25 minutes due to a restriction in our software on the size of video file acquired. In this case, the 30 minute FBI period would use minutes 1-25 only. However, the 30 minute FBI was only used as an indicative overall evaluation (e.g. SI figure S3) because shorter periods of 10 minutes, updated every 1 minute, were adjudged to be representative analysis periods. The system was also tested over periods of 40 minutes at a time. The number, extent and updating frequency of the FBI time periods may be further adapted to suit the particular design of experiments and interventions. Fig. S1 illustrates both the latest FBI for different time periods (Fig. S1(a)) and the history detail of the 10 min and 1 min FBI ( Fig. S1(b)). Fig. S1(a) shows the higher-level overview of FBI as in Fig. 2 of the main paper. Here, the fish is exhibiting normal healthy behavior on the various timescales. At a deeper historical level, the 1 min indicator shown on Fig. S1(b) provides a picture of behavior on a short timescale and shows considerable fluctuations between Normal and Abnormal FBI. The two Abnormal FBI states in this 40 min period are circled on Fig. S1(b). It was observed that the healthy fish unexpectedly had 2 quiescent periods at minutes 33 and 36 (prolonged periods of resting on the tank bottom) which would be more expected from unhealthy fish. By contrast the 10 min FBI remains constant and at maximal rating of Healthy for this fish after the first 12 minutes. At an even deeper level, Fig. S2 below illustrates for a Control fish and a PIT tagged fish, displaying the extremes of the scale or index, the 'raw' data for their Activity and Distance parameters. Fig. 2 (a) and (d) (main paper) shows the FBI results derived from such underlying raw data of Fig S2. The FBI analysis module is currently contained in a separate Microsoft Excel file for ease of use and can provide real-time functionality or alternatively be used to analyze subjects' test files (coordinate logs) retrospectively (post processing). It generates the FBI over periods of the latest 1 minute, 10 minutes, 20 minutes and 30 minutes (with provision also for latest 60 minutes). The FBI is qualitative (with categories Healthy, Ok, Unhealthy, Abnormal) and is currently based on a combination of Distance Swum (DS) and Activity (A). S2 shows raw Activity as zones (1-9) visited over time (with the tank divided into 9 zones covering the area available to the fish) and Distance travelled over time for the Control and for the PIT fish of Fig. 2 over 25 minutes at post-treatment time point (3). It also shows the front (side) and top views of the tank using plots of the locations visited by the fish. These data are used in the derivation of the FBI of Fig. 2 (main). The raw data of Fig. S2(b) shows marked differences between a PIT treated fish and the Control fish. Fig. S2(b) shows typical characteristics of abnormal wellbeing: very low raw Activity, low changes in raw Activity level; low Distance swum with several periods of stationary behavior between 'pulses' of movement; low utilization/exploration of the tank. 2. Evaluation of the system The deeper system components, 10- The 10-minute FBI gives considerable detail on developments in welfare. In period (ii), C8 declined in its normal activity, going through an 'Ok' phase which declined further to Unhealthy and Abnormal for 3 and 4 minutes respectively before recovering through U and O for 2 and 15 minutes respectively. In period (iii) it was again Healthy overall, as expected for an untreated fish. The effect of the fin clip on FC2 was dramatic as the fish immediately went into Abnormal condition throughout periods (ii) and (iii) without recovering. By contrast, FC6 declined over periods (ii) and (iii) but only reached Abnormal for 3 minutes. Otherwise, the fish recovered to Healthy for 7 minutes in period (ii). Overall, it appeared less affected by the treatment than the severely affected FC2. FBI characterizes more gradual variation in behavior over the medium-short term, whereas the longer term 30 min view was far less sensitive to short term fluctuations. S4 shows 10min FBI updated minute by minute. It further illustrates the interplay between 30min and 10min FBI for several behaviors. Sham1 ( Fig. S4(b)), has a 'quieter' period of 12 minutes in period (iii) seen in the consecutive 10min FBI 'Ok' categorizers extending from minute 3 to minute 14. Nevertheless, overall (25min) it is adjudged Healthy. AL1 ( Fig. S4(c)) has similar dips in period (i), which include Unhealthy and Ok but is overall Healthy. In period (iii) it is overall Unhealthy (but not worse than that) after dipping to Abnormal for 13 minutes extending from minute 6 to minute 18. The 10min FBI characterizes more gradual variation in behavior over the medium-short term. The longer term 30 minute view is far less sensitive to short term fluctuations. (An extended version of Figure S4 Validation in Two Dimensions FBI dependence on 3D versus 2D monitoring was investigated since many laboratories may not have the capacity to put video cameras above tanks in a typical zebrafish racking system. This affected distance travelled thus the software was modified to operate with one camera (side view) and FBI was adapted (publically demonstrated at the Blue Planet Aquarium 2015 (www.blueplanetaquarium.com/fish-health-monitor-trial-this-saturday-at-blue-planet). To compare 2D with 3D FBI, the 3D coordinates for the 5 fish of Fig. 2 (main) were also run through the modified 2D FBI system (i.e. 2D FBI utilizing front camera coordinates only). Fifteen videos from experiment 1 were chosen at random and re-assessed using only 2D data. Out of 15 only 2 differed from 3D analysis but were within the Healthy/OK or within Unhealthy/Abnormal categories thus the overall wellbeing status was within healthy and Equipment Setup and Operation The system runs tracking software 15 Six are detailed in the following Table 1 and an example screen of the system in use is shown in Fig. S6. Additional videos include the tracking system in operation and an example of a fish improving to Healthy from Unhealthy (SI Videos 1 and 2). Overall for this timescale: seems Normal, intermittently exploring full tank volume. The following screen example (Fig. S6) shows the fish subject (right, identified with pink circles), its trajectory (center right) and the detailed FBI analyses (left half) and figure S7 demonstrates how the FBI assesses zebrafish from each treatment group 2 h after treatment. Fig S7 presents, for all the subjects in the 5 groups of female zebrafish, mean FBI for the 30 minute timescale at the 3 initial timepoints (pre, 1h, 2h). For clarity, each group is normalized to its mean Pre value. FBI of the Control and Sham groups remained virtually constant and Healthy throughout. FBI for the Fin Clip group, which is most affected, declined by ~50% at timepoint 2 and another ~25% from timepoint 2 to 3, taking this group into the Abnormal FBI category. The other treated groups responded similarly to Fin Clip but with reduced severity. PIT ended at timepoint 3 with FBI ~50% reduced from pre-treatment whereas the least affected group, Acid Lip, has FBI reduced by about 35%. Acid Lip and PIT treated fish reached the Unhealthy category but avoided the Abnormal category. The results from the FBI in Figure S7 reflect the statistically significant results obtained for the Acid Lip 1%, PIT and Fin Clip treatment groups (Table S2) based on individual parameters from the Principle Components Analysis (Figs. 4 and 5). Discussion Behavioural effects of treatment: Female zebrafish were profoundly affected by the invasive procedures employed in the present study with reductions in swimming speed and the amount of the tank explored and an increase in the use of the bottom of the tank. The fin clip and 5/10% acid lip groups had not recovered by the end of the 6 h experiment and their behaviour still differed from pre-treatment behaviour. This prolonged, complicated behavioural change over 6 hours with no recovery indicates that this was not a simple nocifensive reflex 1 but was a substantial modification of the animal's normal behaviour 2 . This response was distinguishable from sham-handled fish whose behaviour did not differ discernibly from controls thus the changes in behaviour are not due to the stress of anaesthesia and handling 1 . These changes were ameliorated by the administration of pain-relieving drugs and thus this information can be used to develop analgesic protocols for fish. Acetic acid: The injection of increasing concentrations of acetic acid provoked a change in behaviour, similar to that observed in a previous studies 3 , which was dose dependent. The injection of 1% acetic acid had only a minor impact on the percentage of tank-explored post 2 and 3h before exploration returned to normal. An increase in concentration up 10%, however, yielded a much greater behavioural response resulting in a significant change across all three behavioural traits. Other studies have used direct observation rather than a behavioural analysis tool but similarly have demonstrated zebrafish reduced their activity after a potentially painful event 3,4 . A reduction in activity and exploration has been observed in zebrafish and trout during painful stimulation 4 and mirror the immobile states and reduced exploration that higher vertebrates can experience in response to a noxious stimulus 5 . Reilly et al. 4 hypothesised that these behaviours may serve to protect the animal from predation when injured by allowing the conservation of energy that may be needed in a fight or flight response and making the individual less conspicuous. PIT tagging. The insertion of PIT tags resulted in a departure from control behaviour 2 hours after implantation and was characterized by a reduction in average speed and tank exploration with a clear preference for the bottom of the tank. The potential for this procedure to cause pain is considerable given the risk of damage to the musculature of the abdominal cavity, yet the addition of extra weight via the PIT tag could add an energetic cost to movement resulting in the observed reduction in activity. Previous studies investigating the impact of tagging on a broad spectrum of species found no effects on critical swimming velocity 6-8 even when the weight of the tags reached 6-12% of the bodyweight of the individual 8 which are well above the relative weight of the PIT tags (2% of zebrafish bodyweight) used in this study. Fin clipping. The fin clip procedure resulted in reductions in average speed, tank exploration and a clear preference for the bottom of the tank after treatment. The fin clip procedure resulted in a significant departure from normal behaviour earlier at 1 h and across more time points suggesting that the impact of the fin clip was more immediate and potentially of greater severity relative to the PIT tag. These changes in behaviour have been observed in previous studies in zebrafish 9.10 . Currently, fin clip is a procedure which is deemed mild severity under EU legislation 11 and is believed to result in mild or acute pain for a few hours but our results show that the responses to fin clipping persists for several hours and as such should be deemed moderately severe. The use of immersion analgesia to alleviate any associated pain would ensure that any pain would be reduced and the procedure was indeed mild. As with the PIT tag procedure, it is possible that the reduction in activity was related to the physical impediment of having 40% less tail fin as opposed to being a complex response to pain. The complete absence of a tail fin in the no-tail strain of zebrafish resulted in a 65% reduction in critical swimming performance 12 , an effect that could account for the observed changes in behaviours linked to activity. If the changes in behaviour are related to the ability to swim with a shorter tail then the administration of analgesics would have no affect; however, certain analgesics at a specific dose helped to ameliorate the behavioural impact of the fin clip thereby confirming the potential for this procedure to be painful. Impact of drug use: The efficacies of four drugs, from three different classes of analgesics (NSAIDs, opioids and local anaesthetics), differed in their ability to prevent the behavioural change induced by the fin clipping in female zebrafish. Lidocaine (5mg/L) was successful analgesic, since it reduced the effect of the fin clip across all behaviours for the duration of the experiment. Lidocaine treated zebrafish exhibited behaviours that consistently aligned with that observed in the control group even though they had been fin clipped. Studies have found lidocaine to be effective in rainbow trout 13 and zebrafish 10 . Together these findings further validate this drug as a promising analgesic in fish as it has now shown effectiveness in alleviating more than one pain type (chemical and mechanical) in both a cyprinid and a salmonid. The second local analgesic tested was not quite as successful: at 6h fish treated with all doses of bupivacaine displayed behaviour akin to fin clip without analgesia since they were significantly different from controls. In mammalian models the local anaesthetic bupivacaine can have a superior duration of activity 14 ; however, this action was not observed here. Instead the highest dose (1 mg/L) did not alleviate the fin clip procedure to the same degree as the lidocaine and can be deemed less effective. Perhaps a higher dose of bupivacaine is required and this should be explored in future studies. The opioid morphine and NSAID flunixin both proved effective and had a clear dose dependent effect with the highest dose in each case leading to the greatest reduction in fin clip mediated behavioural change. Morphine is effective at ameliorating the impact of noxious stimuli across a large spectrum of vertebrates [15][16][17][18] including teleost fish 19 . In this current study, 48mg/L of morphine treated fish behaviour showed average speeds that were similar to control fish. The reverse, however, was observed with the lowest dose of morphine. Although previous studies have largely focused on the injection of morphine, only two have looked at administering morphine via the immersion route in goldfish (Carassius auratus) 20,21 . Jansen and Greene 21 demonstrated the rapid uptake of morphine from water in goldfish, however, Newby et al. 20 could not replicate these results and found a much slower rate of uptake. Despite this slow uptake the high dose of morphine (48mg/L) still resulted in a reduction in pain related behaviours; morphine within the water was therefore hypothesised to act centrally 20 . The results from this current study demonstrate the effectiveness of morphine administered via the immersive route, albeit at a high dose, although future work should examine the uptake kinetics and potential side effects of morphine in zebrafish to help better determine an effective dosage. The highest dose (8mg/L) of the NSAID flunixin meant fish behaviour did not differ over time and was similar to controls except at the 6h time point. The behaviour average speed in the 8mg/L group were similar to that observed in controls and elevated compared to that observed in the fin clip group. The efficacy seen in the highest dose steadily decreased with dose with the behaviour of the 2mg/L flunixin group resembling that of fin clipped fish and average speed was significantly different from controls at 2, 3 and 6h. NSAID's function through the inhibition of the enzymes arachidonate cyclo-oxygenase 1 (COX-1) and 2 (COX-2); functional genes for both of these enzymes are conserved in zebrafish 22 and levels of the NSAID diclofenac as low as 1 μg/L are known to reduce COX expression in fish 23 . This would suggest that flunixin at 8mg/L is not as effective as lidocaine or morphine or that future studies should test higher doses. Conclusion. Taken together it would seem from the present study that lidocaine (5mg/L) administered via immersion prior to treatment is the most effective drug to prevent fin clip induced changes. Morphine was also effective but given the cost and regulatory restrictions in its use it may not be widely adopted. We recommend that analgesia is provided for all invasive procedures that cause tissue damage as it is likely they may give rise to the sensation of pain to ensure good welfare. However, where the analgesic drug itself may confound data collection and justifiably cannot be used then experimenters should wait 24 hours after an invasive procedure before beginning behavioural data collection to allow the zebrafish to recover. After this period it has been shown zebrafish behaviour returns to normal after fin clipping 10 .	2019-06-22T13:41:37.743Z	2019-06-21T00:00:00.000	{ "year": 2019, "sha1": "b68e5283b70a3a36bbfb32283b49ece00602f661", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41598-019-45464-w.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "9136c84d5ef6fff4db2c1a418fa202abe27847ff", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Computer Science", "Medicine" ] }
211522645	pes2o/s2orc	v3-fos-license	Friction and Wear Properties of Ni3Si Alloy with Ti Addition at High Temperatures The tribological properties of Ni3Si alloy were studied at high temperatures. The effect of the addition of Ti was also analyzed. The surface composition was analyzed by Raman spectroscopy. The results showed that the friction coefficient decreased with the increasing temperature, and the wear rate changed slightly from 25 to 400 °C. However, the wear resistance of the alloys decreased sharply at 600 °C, and this was due to the decrease of the high-temperature strength and the severe oxidation of the alloys. Although the oxidation resistance of Ni3Si alloy decreased with Ti addition, the tribological property was improved by the addition of Ti. The Ni3Si alloy with 5% Ti addition had the best wear resistance at high temperatures as compared to pure Ni3Si alloy and with 10% Ti addition, and the wear rates of the alloys were in the order of magnitude of 10−5 mm3/Nm. With the increase of temperature, the wear mechanism of pure Ni3Si alloy transformed from abrasive wear to oxidation wear. As the Ti content increased, the wear mechanisms of the alloys changed from abrasive wear to fatigue wear at low temperature, and oxidation wear and fatigue wear at high temperature. Introduction There is an increasing demand for a wide range of structural materials for high strength, excellent oxidation resistance, low density, good wear, and corrosion resistance. Unfortunately, few kinds of alloys can satisfy all these conditions simultaneously. However, intermetallic compounds have specific mechanical and chemical properties, which can meet these performance requirements. For example, Ni 3 Si alloy possesses an anomalous hardness-temperature relationship (an increase in strength with increasing temperature) and excellent oxidation resistance over a wide range of temperatures [1][2][3][4][5]. In addition, it also exhibits excellent corrosion resistance in sodium chloride and various acidic solutions [6][7][8][9]. Therefore, the alloy is the potential to be used in structural or chemical components, especially in high-temperature fields. Further, it is necessary to study the tribological property of the alloy under harsh environmental conditions. Bi et al. fabricated the Ni 3 Si alloy with Cr addition by combustion synthesis and the tribological properties of the alloy were investigated under high temperature, sulfuric acid solution, dry and water conditions [10][11][12][13]. The results showed that the wear rates of the alloy were in the magnitude of 10 −5 mm 3 /Nm at the temperature below 800 • C; similarly, the wear rates were in the order of 10 −5 mm 3 /m under low and moderate loads both in sulfuric acid solution and water. This indicated that Ni 3 Si alloy possessed excellent wear resistance under these conditions. Niu et al. investigated the tribological properties of Ni 3 Si alloy with Ti addition under vacuum and seawater conditions [14,15]. These results demonstrated that the alloy possessed lower friction coefficient and wear rate in low vacuum condition as compared to high vacuum and air conditions; the tribological properties of Ni 3 Si alloys with different Ti addition are better than that of Ti6Al4V alloy under seawater condition; in addition, the counterpart materials had a significant effect on the tribological behaviors of the alloys, and the order of friction coefficients was Ni 3 Si-Si 3 N 4 > Ni 3 Si-316SS > Ni 3 Si-Al 2 O 3 . Gui et al. studied the tribological properties of NiMo/Mo 2 Ni 3 Si intermetallic composites under dry sliding wear test conditions at room temperature, and found that the wear mass losses and the friction coefficients of the composite were considerably lower than those of the comparison test materials, the austenitic stainless steel 1Cr18Ni9Ti, and hardened 0.45%C steel [16]. Wang et al. observed that Moss-toughened Mo 2 Ni 3 Si alloys exhibited excellent wear resistance and low friction coefficient under dry sliding conditions [17]. Especially under high load, the wear resistance of the alloy was 10 times more than that of hardened 0.45%C steel. Wang et al. also investigated the tribological properties of Ti 2 Ni 3 Si/NiTi intermetallic composite coating and the high-temperature sliding wear resistance of the γ-toughened Cr 13 Ni 5 Si 2 alloy [18,19]. The Ti 2 Ni 3 Si/NiTi coating had excellent abrasive and adhesive wear resistance under dry sliding conditions, and the wear loss reduced by more than 15 times when compared with the hardened 0.45%C carbon steel and the hardened high-carbon low-alloy tool-steel 1.0%C-1.5%Cr. Wear volume loss of the Cr 13 Ni 5 Si 2 alloy was dramatically lower than that of the austenitic steel 1Cr18Ni9Ti and was even slightly decreased with the increasing test temperature. From the abovementioned work, the study on the tribological properties of Ni 3 Si alloy mainly focused on the effect of alloying elements of titanium and chromium. However, with the addition of Cr, Ni 3 Si alloy was apt to form a brittle phase, and this led to increased strength and decreased toughness [20]. On the contrary, the strength and ductility of Ni 3 Si alloy increased with Ti addition [2], but the high-temperature tribological properties of the alloy were rarely studied. The reason might be that the oxidation resistance of Ni 3 Si alloy declined slightly with the decrease of Si content [21]. It is noted that a high content of oxides is conducive to reduce friction and wear [22,23]. Therefore, the research of high-temperature tribological properties of Ni-Si-Ti alloy will provide a comprehensive understanding of the alloy, which is beneficial to the application of the alloy in high-temperature service condition. In the present work, the wear resistance and friction coefficient of Ni 3 Si alloy with Ti addition were evaluated at high temperatures. The effects of the temperature and the content of Ti were discussed, and the corresponding wear mechanisms were analyzed. Materials and Methods The Ni 3 Si alloys with different Ti addition were prepared by mechanical alloying and vacuum hot-pressing sintering. The preparation process is described in detail elsewhere [24]. The tribological tests were carried out by a ball-on-disk tribometer with the alloys and bearing steel balls as a counter material. The alloy disk with a dimension of 18 × 18 × 3 mm 3 was rotated against a commercial Si 3 N 4 ceramic ball (diameter of 5 mm, hardness of 15 GPa, roughness of 0.02 µm). Before each experiment, the specimens were ground with SiC abrasive paper and polished with diamond abrasive to Ra ≤ 0.06 µm, followed by ultrasonic cleaning in an acetone solution. The applied loads were 10 N, and the sliding speed was 0.10 m/s. The total sliding time was 20 min. The test temperatures were 25, 200, 400, and 600 • C. The wear rate was calculated by the expression: W = SC/NL, where S is the cross-sectional area, C is wear track perimeter, N is the applied load, and L is sliding distance. The cross-sectional profile of the worn surface was measured using a confocal laser scanning microscope (LEXT OLS4000, Tokyo, Japan), and the area was calculated automatically by the equipment. All the tests were carried out at least three times under identical conditions. A scanning electron microscope (SEM, JSM-5600LV, Tokyo, Japan) equipped with energy dispersive spectroscopy (EDS) (Oxford Instruments, Abingdon, UK) was used to study the morphology characteristics and chemical composition of the worn surfaces. The SERS measurement was carried out with a Renishaw InVia Raman microscope (Gloucestershire, UK). Tribological Behaviors of the Alloys The typical curves of friction coefficients (COFs) as a function of time are shown in Figure 1. It was clear that the COFs became stable after a short running-in period, and the value was higher at room temperature as compared to high temperatures. The fluctuations of the friction coefficient curves were relatively low at high temperatures, the reason for this might be that the oxides played a critical role in reducing friction and maintaining the friction force stable, which formed during the sliding process at high temperature. The friction coefficient of the alloy with 10% Ti addition was lowest at 400 • C. Tribological Behaviors of the Alloys The typical curves of friction coefficients (COFs) as a function of time are shown in Figure 1. It was clear that the COFs became stable after a short running-in period, and the value was higher at room temperature as compared to high temperatures. The fluctuations of the friction coefficient curves were relatively low at high temperatures, the reason for this might be that the oxides played a critical role in reducing friction and maintaining the friction force stable, which formed during the sliding process at high temperature. The friction coefficient of the alloy with 10% Ti addition was lowest at 400 °C. The friction coefficient as a function of temperature for Ni3Si alloy is given in Figure 2. As the temperature increased, the friction coefficient of pure Ni3Si alloy showed a descending trend. This could be related to the good high-temperature mechanical properties and the oxidation of the alloy at high temperatures. With the addition of 5% Ti, the alloy had a low friction coefficient at room temperature, and a higher friction coefficient at 200 °C. Then, the friction coefficient decreased with the temperature rising. The Ni3Si alloy with 10% Ti addition exhibited an obvious decrease in friction coefficient from room temperature to 400 °C, and a slight increase in friction coefficient at 600 °C. The reverse trend between the two alloys was due to the different content of Ti addition. The friction coefficient as a function of temperature for Ni 3 Si alloy is given in Figure 2. As the temperature increased, the friction coefficient of pure Ni 3 Si alloy showed a descending trend. This could be related to the good high-temperature mechanical properties and the oxidation of the alloy at high temperatures. With the addition of 5% Ti, the alloy had a low friction coefficient at room temperature, and a higher friction coefficient at 200 • C. Then, the friction coefficient decreased with the temperature rising. The Ni 3 Si alloy with 10% Ti addition exhibited an obvious decrease in friction coefficient from room temperature to 400 • C, and a slight increase in friction coefficient at 600 • C. The reverse trend between the two alloys was due to the different content of Ti addition. The wear rate as a function of temperature for Ni 3 Si alloy is shown in Figure 3. The wear rate of pure Ni 3 Si alloy decreased slightly in the temperature range from 25 to 400 • C. The alloy with 10% Ti addition had the lowest wear rate at room temperature, and a little lower wear rate at the medium temperature as compared to pure Ni 3 Si alloy. The alloy with the addition of 5% Ti exhibited the lowest wear rate at high temperatures, and showed the optimal wear resistance among the tested materials. When the temperature rose from 400 to 600 • C, the wear rates of the alloys all increased sharply, especially pure Ni 3 Si alloy. It was noted that the wear rates of the alloys were in the magnitude of 10 −5 mm 3 /Nm. could be related to the good high-temperature mechanical properties and the oxidation of the alloy at high temperatures. With the addition of 5% Ti, the alloy had a low friction coefficient at room temperature, and a higher friction coefficient at 200 °C. Then, the friction coefficient decreased with the temperature rising. The Ni3Si alloy with 10% Ti addition exhibited an obvious decrease in friction coefficient from room temperature to 400 °C, and a slight increase in friction coefficient at 600 °C. The reverse trend between the two alloys was due to the different content of Ti addition. The wear rate as a function of temperature for Ni3Si alloy is shown in Figure 3. The wear rate of pure Ni3Si alloy decreased slightly in the temperature range from 25 to 400 °C. The alloy with 10% Ti addition had the lowest wear rate at room temperature, and a little lower wear rate at the medium temperature as compared to pure Ni3Si alloy. The alloy with the addition of 5% Ti exhibited the lowest wear rate at high temperatures, and showed the optimal wear resistance among the tested materials. When the temperature rose from 400 to 600 °C, the wear rates of the alloys all increased sharply, especially pure Ni3Si alloy. It was noted that the wear rates of the alloys were in the magnitude of 10 −5 mm 3 /Nm. Wear Mechanisms of the Alloys The surface topographies of the wear tracks of pure Ni3Si alloy are displayed in Figure 4. It was clear that there were numerous scratches and furrows on the worn surface in the temperature range from 25 to 200 °C (Figure 4a,b). This indicated that the wear mechanisms of pure Ni3Si alloy were abrasive wear at these temperatures. When the temperature was up to 400 °C, some shallow grooves, as well as fine wear debris existed on the worn surface ( Figure 4c). The main wear mechanism was abrasive wear, and accompanied with oxidation wear at the temperature. At 600 °C, the worn surface was covered with a compact oxide film, and the composition of the oxide was 51.83Ni-14.83Si-33.34O. However, the oxide film was discontinuous with granular structure, and unable to effectively protect the sublayer. Therefore, the alloy suffered severe oxidation wear, and the wear rate of the alloy rose rapidly at this temperature. In addition, for pure Ni3Si alloy, the yield strength was highest at 400 °C, and then decreased significantly [25]. On that basis, the wear rate and the friction coefficient of the alloy decreased from 25 to 400 °C. At the temperature of 600 °C, the oxidation of the alloy occurred obviously, and the granular oxide layer formed and covered on the entire surface ( Figure 4d). The oxide layer did not form a glaze layer on the worn surface and tended to peel off. Combined with the decreased strength above 400 °C, the friction coefficient of the alloy decreased slightly, and the wear rate had a dramatic increase. Wear Mechanisms of the Alloys The surface topographies of the wear tracks of pure Ni 3 Si alloy are displayed in Figure 4. It was clear that there were numerous scratches and furrows on the worn surface in the temperature range from 25 to 200 • C (Figure 4a,b). This indicated that the wear mechanisms of pure Ni 3 Si alloy were abrasive wear at these temperatures. When the temperature was up to 400 • C, some shallow grooves, as well as fine wear debris existed on the worn surface ( Figure 4c). The main wear mechanism was abrasive wear, and accompanied with oxidation wear at the temperature. At 600 • C, the worn surface was covered with a compact oxide film, and the composition of the oxide was 51.83Ni-14.83Si-33.34O. However, the oxide film was discontinuous with granular structure, and unable to effectively protect the sublayer. Therefore, the alloy suffered severe oxidation wear, and the wear rate of the alloy rose rapidly at this temperature. In addition, for pure Ni 3 Si alloy, the yield strength was highest at 400 • C, and then decreased significantly [25]. On that basis, the wear rate and the friction coefficient of the alloy decreased from 25 to 400 • C. At the temperature of 600 • C, the oxidation of the alloy occurred obviously, and the granular oxide layer formed and covered on the entire surface (Figure 4d). The oxide layer Materials 2020, 13, 982 5 of 11 did not form a glaze layer on the worn surface and tended to peel off. Combined with the decreased strength above 400 • C, the friction coefficient of the alloy decreased slightly, and the wear rate had a dramatic increase. The morphologies of the wear scars of Ni3Si alloy with 5% Ti addition are shown in Figure 5. The worn surface of the alloy was divided into two regions at room temperature: A contact region and a wear region (Figure 5a). The compositions of the contact region and the wear region were 58.4Ni-16.3Si-14.2Ti-10.0O and 68.2Ni-26.2Si-5.6O, respectively. This was because of the harder phase Ni3Ti (about 700 HV) formed during the sintering process, which was able to support the applied load effectively [24]. In contrast, the Si-rich region (the wear region, about 500 HV) had lower hardness, and was worn out firstly. As a result, the wear resistance of the alloy was improved as compared to pure Ni3Si alloy, and the real area of contact was smaller during the sliding process. The alloy showed a relatively low friction coefficient and wear rate at room temperature, and the wear mechanism was slightly abrasive wear. As the temperature rose to 200 °C, the worn surface was covered with some oxides besides some shallow grooves (as shown in Figure 5b). The composition of the oxide was 30.7Ni-14.8Si-4.2Ti-50.4O, Ti content is relatively high. Along with the oxidation of Ti-rich region, the real area of contact increased, and so the alloy had a high friction coefficient at this temperature. Due to the good high-temperature mechanical property of Ni3Si alloy and the compacted oxide wear debris, the wear rate of the alloy was lower at 200 °C. The wear mechanism was abrasive wear and oxidation wear. When the temperature reached 400 °C, the worn surface was smooth, and covered with oxides in localized areas (shown in Figure 5c). Like pure Ni3Si alloy, the smooth surface was attributed to the highest strength at 400 °C, and the difference is the formation of the oxide layer. This indicated that although the oxidation resistance of the alloy with 5% Ti addition decreases, the wear rate of the alloy was lower. The wear mechanism was oxidation wear at the temperature. At 600 °C, the oxide layer formed (Figure 5d), which composition was 30.4Ni-9.7Si-2.3Ti-57.6O. The uncovered area had been worn out and was re-oxidized during the sliding process. The EDS result showed that the composition of this area is 57.5Ni-19.3Si-1.4Ti-21.8O, which had a low oxidation degree. The significant oxidation and wear resulted in a low friction coefficient and high wear rate simultaneously as compared to that at 400 °C. It was worth pointing out that the morphology of the worn surface of The morphologies of the wear scars of Ni 3 Si alloy with 5% Ti addition are shown in Figure 5. The worn surface of the alloy was divided into two regions at room temperature: A contact region and a wear region (Figure 5a). The compositions of the contact region and the wear region were 58.4Ni-16.3Si-14.2Ti-10.0O and 68.2Ni-26.2Si-5.6O, respectively. This was because of the harder phase Ni 3 Ti (about 700 HV) formed during the sintering process, which was able to support the applied load effectively [24]. In contrast, the Si-rich region (the wear region, about 500 HV) had lower hardness, and was worn out firstly. As a result, the wear resistance of the alloy was improved as compared to pure Ni 3 Si alloy, and the real area of contact was smaller during the sliding process. The alloy showed a relatively low friction coefficient and wear rate at room temperature, and the wear mechanism was slightly abrasive wear. As the temperature rose to 200 • C, the worn surface was covered with some oxides besides some shallow grooves (as shown in Figure 5b). The composition of the oxide was 30.7Ni-14.8Si-4.2Ti-50.4O, Ti content is relatively high. Along with the oxidation of Ti-rich region, the real area of contact increased, and so the alloy had a high friction coefficient at this temperature. Due to the good high-temperature mechanical property of Ni 3 Si alloy and the compacted oxide wear debris, the wear rate of the alloy was lower at 200 • C. The wear mechanism was abrasive wear and oxidation wear. It was well-known that a lot of oxides were compressed and smeared into the surface during the sliding process, and this continued until the sintered surface layer of thickness was sufficient to prevent further disruption of the alloy surface [26]. Although the oxidation resistance decreased with the addition of Ti, the tribological property of the alloy with Ti addition was improved as compared to pure Ni3Si alloy. Therefore, the friction coefficient and wear rate of the alloy with 5% Ti addition were lower than those of pure Ni3Si alloy at this temperature. The morphologies of the wear scars of Ni3Si alloy with 10% Ti addition are given in Figure 6. The compositions of A and B regions were 61.2Ni-6.3Si-24.4Ti-8.0O and 31.0Ni-7.2Si-2.8Ti-59.0O, respectively (shown in Figure 6a). With the increase of Ti content, the area of the Ti-rich region (A region) increased, and the morphology of the region was relatively smooth without obvious grooves. The smoother surface caused the lower wear rate and the higher real area of contact. Therefore, Ni3Si alloy with 10% Ti addition showed a high coefficient of friction and a low wear rate simultaneously at room temperature. It was noted that the worn surface appeared some wear debris and cracks, and the fluctuation of the friction coefficient might be attributed to the interaction of wear debris and localized fracture of the tribolayer [27,28]. The main wear mechanism was oxidation wear, along with fatigue wear at room temperature. The elementary compositions of C and D regions were 62.3Ni-14.1Si-13.4Ti-10.3O and 27.6Ni-7.7Si-3.4Ti-61.4O, respectively (shown in Figure 6b). A large number of fine oxide particles stacked on the worn surface, and played a role in reducing friction. Although the Ti-rich region was smooth, the area of the flake pits was larger, and thus the alloy with 10% Ti addition had a moderate wear rate among the tested alloys at 200 °C. As the temperature rose further, the oxide layer formed on the worn surface, and the uncovered area was smooth, which was similar to that of pure Ni3Si alloy and with 5% Ti addition at the same temperature. The compact oxide layer and the smooth surface maintained the friction coefficient and wear rate at a low level at 400 °C (Figure 6c). When the temperature reached 600 °C, the oxidation was further accelerated, the oxides were swept out of the wear track before the oxide layer could When the temperature reached 400 • C, the worn surface was smooth, and covered with oxides in localized areas (shown in Figure 5c). Like pure Ni 3 Si alloy, the smooth surface was attributed to the highest strength at 400 • C, and the difference is the formation of the oxide layer. This indicated that although the oxidation resistance of the alloy with 5% Ti addition decreases, the wear rate of the alloy was lower. The wear mechanism was oxidation wear at the temperature. At 600 • C, the oxide layer formed (Figure 5d), which composition was 30.4Ni-9.7Si-2.3Ti-57.6O. The uncovered area had been worn out and was re-oxidized during the sliding process. The EDS result showed that the composition of this area is 57.5Ni-19.3Si-1.4Ti-21.8O, which had a low oxidation degree. The significant oxidation and wear resulted in a low friction coefficient and high wear rate simultaneously as compared to that at 400 • C. It was worth pointing out that the morphology of the worn surface of the alloy with 10% Ti addition was quite different from that of pure Ni 3 Si alloy at 600 • C. The continuous dense oxide layer (Figure 4d) showed the better oxidation resistance of pure Ni 3 Si alloy than the alloy with Ti addition. However, the better oxidation resistance was, the less oxides formed. It was well-known that a lot of oxides were compressed and smeared into the surface during the sliding process, and this continued until the sintered surface layer of thickness was sufficient to prevent further disruption of the alloy surface [26]. Although the oxidation resistance decreased with the addition of Ti, the tribological property of the alloy with Ti addition was improved as compared to pure Ni 3 Si alloy. Therefore, the friction coefficient and wear rate of the alloy with 5% Ti addition were lower than those of pure Ni 3 Si alloy at this temperature. The morphologies of the wear scars of Ni 3 Si alloy with 10% Ti addition are given in Figure 6. The compositions of A and B regions were 61.2Ni-6.3Si-24.4Ti-8.0O and 31.0Ni-7.2Si-2.8Ti-59.0O, respectively (shown in Figure 6a). With the increase of Ti content, the area of the Ti-rich region (A region) increased, and the morphology of the region was relatively smooth without obvious grooves. The smoother surface caused the lower wear rate and the higher real area of contact. Therefore, Ni 3 Si alloy with 10% Ti addition showed a high coefficient of friction and a low wear rate Materials 2020, 13, 982 7 of 11 simultaneously at room temperature. It was noted that the worn surface appeared some wear debris and cracks, and the fluctuation of the friction coefficient might be attributed to the interaction of wear debris and localized fracture of the tribolayer [27,28]. The main wear mechanism was oxidation wear, along with fatigue wear at room temperature. The elementary compositions of C and D regions were 62.3Ni-14.1Si-13.4Ti-10.3O and 27.6Ni-7.7Si-3.4Ti-61.4O, respectively (shown in Figure 6b). A large number of fine oxide particles stacked on the worn surface, and played a role in reducing friction. Although the Ti-rich region was smooth, the area of the flake pits was larger, and thus the alloy with 10% Ti addition had a moderate wear rate among the tested alloys at 200 • C. Materials 2020, 13, x FOR PEER REVIEW 7 of 11 form on the worn surface (Figure 6d). The presence of oxide particles was helpful to have a low level of friction coefficient. At the same time, the intense oxidation brought severe delamination for the worn surface. Further, the wear rate of the alloy with 10% Ti addition increased remarkably at 600 °C. Raman Spectrum Analyses of the Alloys The Raman spectra of the alloys at 600 °C are shown in Figure 7. It was observed that all the Raman spectra of the alloys had a broad intense Raman band in the 700-400 cm −l region. The bands in this region are associated with the presence of bridging oxygens (Si-O-Si) in the structure [29]. This indicated that the worn surface was heavily oxidized at high temperature. However, the compositions of the surface of the alloys were different. For pure Ni3Si alloy, the Raman peak was located at 560 cm −1 , and the band was attributed to the Si-O-Si symmetric stretching vibration mode of Q 3 species [30]. In the meantime, the band also originated from the LO mode of nickel oxide [31]. An analogous situation occurred in the high-frequency region. The 1100 cm −1 band was assigned to Si-O stretching vibration modes of Q 3 species [30], and the characteristic peak was also attributed to 2LO mode of NiO [31]. According to the surface morphology of pure Ni3Si alloy (Figure 4d), the discontinuous layer consisted of spherical particles, and the morphology was in accord with that of NiO [32]. This meant that NiO indeed existed on the worn surface after the friction test, but the oxidation of Ni3Si alloy was prone to form an external NiO layer and an internal SiO2 layer [1,21]. In other words, the observed worn surface was the newly oxidized layer, and the entire surface suffered serious oxidation wear. Therefore, the wear rate of pure Ni3Si alloy increased sharply, and was the highest at 600 °C. With the addition of Ti, the Raman spectra showed peaks located at 548 cm −1 , and the band could be assigned to the Ti-O vibrations and the Si-O-Si bending vibrations [33,34]. In addition, a new band observed at 1016 cm −1 was attributed to the Si-O stretching of Q 2 units [35]. This presented that the As the temperature rose further, the oxide layer formed on the worn surface, and the uncovered area was smooth, which was similar to that of pure Ni 3 Si alloy and with 5% Ti addition at the same temperature. The compact oxide layer and the smooth surface maintained the friction coefficient and wear rate at a low level at 400 • C (Figure 6c). When the temperature reached 600 • C, the oxidation was further accelerated, the oxides were swept out of the wear track before the oxide layer could form on the worn surface (Figure 6d). The presence of oxide particles was helpful to have a low level of friction coefficient. At the same time, the intense oxidation brought severe delamination for the worn surface. Further, the wear rate of the alloy with 10% Ti addition increased remarkably at 600 • C. Raman Spectrum Analyses of the Alloys The Raman spectra of the alloys at 600 • C are shown in Figure 7. It was observed that all the Raman spectra of the alloys had a broad intense Raman band in the 700-400 cm −l region. The bands in this region are associated with the presence of bridging oxygens (Si-O-Si) in the structure [29]. This indicated that the worn surface was heavily oxidized at high temperature. However, the compositions of the surface of the alloys were different. For pure Ni 3 Si alloy, the Raman peak was located at 560 cm −1 , and the band was attributed to the Si-O-Si symmetric stretching vibration mode of Q 3 species [30]. In the meantime, the band also originated from the LO mode of nickel oxide [31]. An analogous situation occurred in the high-frequency region. The 1100 cm −1 band was assigned to Si-O stretching vibration modes of Q 3 species [30], and the characteristic peak was also attributed to 2LO mode of NiO [31]. According to the surface morphology of pure Ni 3 Si alloy (Figure 4d), the discontinuous layer consisted of spherical particles, and the morphology was in accord with that of NiO [32]. This meant that NiO indeed existed on the worn surface after the friction test, but the oxidation of Ni 3 Si alloy was prone to form an external NiO layer and an internal SiO 2 layer [1,21]. In other words, the observed worn surface was the newly oxidized layer, and the entire surface suffered serious oxidation wear. Therefore, the wear rate of pure Ni 3 Si alloy increased sharply, and was the highest at 600 • C. Materials 2020, 13, x FOR PEER REVIEW 8 of 11 coefficient and wear rate among the alloys at 600 °C. The alloy with the addition of 10% Ti had a similar composition as the alloy with 5% Ti addition. However, the intensities of the peaks were lower, and this could result from the delamination of the worn surface. The delamination led to a higher wear rate, but the medium wear rate was acquired by the presence of a large amount of oxide debris on the worn surface ( Figure 6d). Meanwhile, the accumulated oxides acted as a lubricant in reducing the friction, and the alloy with 10% Ti addition exhibited medium friction coefficient at 600 °C. The alloy with 5% Ti addition had the lowest wear rate among the alloys at high temperatures, and the Raman spectra of the worn surfaces of the alloy at different temperatures are shown in Figure 8. The Raman spectra of the alloy at 200 °C was not given due to the similar curves between the temperature of 25 and 200 °C. It was clear that the main composition of the worn surface of the alloy was still silicon oxide at room temperature, which was similar to that at high temperature (600 °C). The obvious difference was that the peak intensity of the high-frequency region was relatively higher than that of the low-frequency region. It was well-known that the bands in the high-frequency region associated with the stretching vibration mode of nonbridging oxygens, and the bands in the lowfrequency region associated with the vibration of bridging oxygens. This indicated that the oxidation occurred only at the surface, and there was little bonding between the oxidation units (silicon-oxygen tetrahedrons) on account of the low content of bridging oxygens. Therefore, the degree of the oxidation of the alloy was relatively low at room temperature. When the temperature was up to 400 °C, the band shifted to a lower wavenumber from 560 to 527 cm −1 in the low-frequency region. The decreasing shift of the band could be attributed to an increase in the average Si-O-Si angle that occurred as a result of silicate polymerization [29,36]. Moreover, the peak intensity of the low-frequency region was significantly higher than that of the high-frequency region. The good high-temperature strength of the alloy provided the desired wear resistance, and the alloy was hard to wear at 400 °C. The species of oxides were squeezed out and accumulated on the surface during the friction process. Further, the strong bonding between the oxides was achieved by the formation of bridging oxygens, and thus the relative intensity of the bands was much higher in the low-frequency region than that in the high-frequency region. As the temperature increased further, the peak intensities of the alloy were high both in the low-frequency and the high-frequency regions. The reason for this was that the severe oxidation was beneficial to With the addition of Ti, the Raman spectra showed peaks located at 548 cm −1 , and the band could be assigned to the Ti-O vibrations and the Si-O-Si bending vibrations [33,34]. In addition, a new band observed at 1016 cm −1 was attributed to the Si-O stretching of Q 2 units [35]. This presented that the oxidation of the alloy with Ti addition was more severe than that of pure Ni 3 Si alloy. The oxide layer formed on the surface of the alloy with 5% Ti addition, and thus the alloy had the lowest friction coefficient and wear rate among the alloys at 600 • C. The alloy with the addition of 10% Ti had a similar composition as the alloy with 5% Ti addition. However, the intensities of the peaks were lower, and this could result from the delamination of the worn surface. The delamination led to a higher wear rate, but the medium wear rate was acquired by the presence of a large amount of oxide debris on the worn surface ( Figure 6d). Meanwhile, the accumulated oxides acted as a lubricant in reducing the friction, and the alloy with 10% Ti addition exhibited medium friction coefficient at 600 • C. The alloy with 5% Ti addition had the lowest wear rate among the alloys at high temperatures, and the Raman spectra of the worn surfaces of the alloy at different temperatures are shown in Figure 8. The Raman spectra of the alloy at 200 • C was not given due to the similar curves between the temperature of 25 and 200 • C. It was clear that the main composition of the worn surface of the alloy was still silicon oxide at room temperature, which was similar to that at high temperature (600 • C). The obvious difference was that the peak intensity of the high-frequency region was relatively higher than that of the low-frequency region. It was well-known that the bands in the high-frequency region associated with the stretching vibration mode of nonbridging oxygens, and the bands in the low-frequency region associated with the vibration of bridging oxygens. This indicated that the oxidation occurred only at the surface, and there was little bonding between the oxidation units (silicon-oxygen tetrahedrons) on account of the low content of bridging oxygens. Therefore, the degree of the oxidation of the alloy was relatively low at room temperature. Materials 2020, 13, x FOR PEER REVIEW 9 of 11 the bonding between the silicon-oxygen tetrahedrons, and the re-oxidized regions resulted in varying degrees of oxidation of the surface. Finally, the alloy with 5% Ti addition exhibited the best tribological property at 600 °C as compared to the other alloys. Conclusions The effect of the addition of Ti on tribological behaviors of Ni3Si alloy has been investigated under high temperatures. The main conclusions are summarized as follows:  The friction coefficient of Ni3Si alloys showed a declining trend with the increase of temperature. However, the reverse trend was observed between the alloys with Ti addition, and this was due to the different contents of Ti.  The wear rates of the alloys had a little change in the temperature range of 25-400 °C, but the wear rates increased sharply at 600 °C. There were two reasons for this: First, the yield strength reached a maximum at 400 °C, and then decreased with the increase of temperature; second, the severe oxidation occurred at 600 °C. The Ni3Si alloy with 5% Ti addition showed the best wear resistance at high temperatures as compared to pure Ni3Si alloy and with 10% Ti addition. The wear rates of the tested alloys were in the magnitude of 10 −5 mm 3 /m.  The wear mechanism of pure Ni3Si alloy was abrasive wear at low temperature, and oxidation wear at high temperature. When the addition of Ti content increased, the wear mechanisms of the alloys changed from abrasive wear to fatigue wear at low temperature, and oxidation wear and fatigue wear at high temperature. Conflicts of Interest: The authors declare no conflict of interest. When the temperature was up to 400 • C, the band shifted to a lower wavenumber from 560 to 527 cm −1 in the low-frequency region. The decreasing shift of the band could be attributed to an increase in the average Si-O-Si angle that occurred as a result of silicate polymerization [29,36]. Moreover, the peak intensity of the low-frequency region was significantly higher than that of the high-frequency region. The good high-temperature strength of the alloy provided the desired wear resistance, and the alloy was hard to wear at 400 • C. The species of oxides were squeezed out and accumulated on the surface during the friction process. Further, the strong bonding between the oxides was achieved by the formation of bridging oxygens, and thus the relative intensity of the bands was much higher in the low-frequency region than that in the high-frequency region. As the temperature increased further, the peak intensities of the alloy were high both in the low-frequency and the high-frequency regions. The reason for this was that the severe oxidation was beneficial to the bonding between the silicon-oxygen tetrahedrons, and the re-oxidized regions resulted in varying degrees of oxidation of the surface. Finally, the alloy with 5% Ti addition exhibited the best tribological property at 600 • C as compared to the other alloys. Conclusions The effect of the addition of Ti on tribological behaviors of Ni 3 Si alloy has been investigated under high temperatures. The main conclusions are summarized as follows: • The friction coefficient of Ni 3 Si alloys showed a declining trend with the increase of temperature. However, the reverse trend was observed between the alloys with Ti addition, and this was due to the different contents of Ti. • The wear rates of the alloys had a little change in the temperature range of 25-400 • C, but the wear rates increased sharply at 600 • C. There were two reasons for this: First, the yield strength reached a maximum at 400 • C, and then decreased with the increase of temperature; second, the severe oxidation occurred at 600 • C. The Ni 3 Si alloy with 5% Ti addition showed the best wear resistance at high temperatures as compared to pure Ni 3 Si alloy and with 10% Ti addition. The wear rates of the tested alloys were in the magnitude of 10 −5 mm 3 /m. • The wear mechanism of pure Ni 3 Si alloy was abrasive wear at low temperature, and oxidation wear at high temperature. When the addition of Ti content increased, the wear mechanisms of the alloys changed from abrasive wear to fatigue wear at low temperature, and oxidation wear and fatigue wear at high temperature. Conflicts of Interest: The authors declare no conflict of interest.	2020-02-27T09:33:18.283Z	2020-02-01T00:00:00.000	{ "year": 2020, "sha1": "cf5fa1983446ba897c1db8560842ee392a1cb91b", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1996-1944/13/4/982/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "00a7883c4c2783732101fc397e63747e1c1341ee", "s2fieldsofstudy": [ "Materials Science" ], "extfieldsofstudy": [ "Medicine", "Materials Science" ] }
119129942	pes2o/s2orc	v3-fos-license	Weak dual equivalence for polynomials We use dual equivalence to give a short, combinatorial proof that Stanley symmetric functions are Schur positive. We introduce weak dual equivalence, and use it to give a short, combinatorial proof that Schubert polynomials are key positive. To demonstrate further the utility of this new tool, we use weak dual equivalence to prove a nonnegative Littlewood--Richardson rule for the key expansion of the product of a key polynomial and a Schur polynomial, and to introduce skew key polynomials that, when skewed by a partition, expand nonnegatively in the key basis. Introduction Schur functions enjoy deep connections with representation theory and algebraic geometry. The quintessential problem of proving that a given function expands nonnegatively in the Schur basis arises because these Schur coefficients enumerate multiplicities of irreducible components or dimensions of algebraic varieties. In earlier work, the author developed a general framework, called dual equivalence, for proving that a given function is symmetric and Schur positive [Ass15]. At its core, the method imposes a rigid structure on the set of combinatorial objects that generate the given function that ensures there is a weight-preserving bijection with standard Young tableaux, the latter of which generate Schur functions. In this paper, we begin with a new application of dual equivalence to establish that Stanley symmetric functions [Sta84], introduced by Stanley as a tool to enumerate reduced expressions 2. Schur positivity of Stanley symmetric functions 2.1. Schur functions. Let N and P denote the sets of nonnegative and positive integers, respectively. We use letters a, b, c to denote weak compositions of length n, i.e. sequences in N n , letters α, β, γ to denote strong compositions, i.e. sequences in P k for some k, and λ, µ, ν to denote partitions, i.e. weakly decreasing sequences in P k for some k. Given a weak composition a, let flat(a) denote the strong composition obtained by removing all zero parts. Given a strong composition α, let sort(α) denote the partition obtained by rearranging the parts of α into weakly decreasing order, and extend this to weak compositions by sort(a) = sort(flat(a)). Given a weak composition a, we let x a denote the monomial x a 1 1 · · · x an n , and we use the notation X to denote the infinite set of variables {x 1 , x 2 , . . .} used for functions. For our discussion of symmetric functions, we defer to the beautiful exposition in Macdonald [Mac95], though we use coordinate notation (French) as opposed to matrix notation (English). The Young diagram of a partition λ, denoted by D(λ), is the diagram with λ i unit cells left justified in row i. For example, the Young diagram for (5, 4, 4, 1) is given in Figure 1. The Schur functions, indexed by partitions, form an important basis for symmetric functions. They arise in many contexts, including as irreducible characters for the general linear group and as polynomial representatives for Schubert cycles of the Grassmannian. For our purposes, we define them combinatorially as the quasisymmetric generating function for standard Young tableaux. Definition 2.1 ( [Ges84]). For α a strong composition, the fundamental quasisymmetric function F α is given by where the sum is over weak compositions b whose flattening refines α. A standard Young tableau is a bijective filling of a Young diagram with entries {1, 2, . . . , n} such that entries increase along rows and up columns. Let SYT(λ) denote the set of standard Young tableaux of shape λ. For example, Figure 2 shows the standard Young tableaux of shape (3, 2). For a standard Young tableau T , say that i is a descent of T if i+1 lies weakly left of (equivalently, strictly above) i. The descent composition of T , denoted by Des(T ), is the strong composition given by maximal length runs between descents. For example, see Figure 2. (2, 2, 1) (1, 3, 1) (1, 2, 2) (2, 3) The following definition for a Schur function follows from the classical one (see [Mac95]) by a result due to Gessel [Ges84]. Definition 2.2. For λ a partition, the Schur function s λ is given by F Des(T ) (X). The reverse row reading word of a Young tableau T is obtained by reading the enties right to left along the top row, then right to left for the next row down, and so on. We say that a standard Young tableau is super-standard if its reverse row reading word is the reverse of the identity. For examples, see Figure 3. Proof. Clearly Y is unique, if it exists, and it can be constructed by placing entries 1, . . . , λ 1 in the bottom row, λ 1 + 1, . . . , λ 1 + λ 2 in the next row up, and so on. Then i will be a descent of Y precisely for i = λ 1 , λ 1 + λ 2 , . . ., giving Des(Y ) = λ. Conversely, if Des(T ) = λ, then T must have descents at i precisely for i = λ 1 , λ 1 + λ 2 , . . .. Therefore the entries 1, 2, . . . , λ 1 must form a horizontal strip, and so must fill the bottom row of λ. Similarly, λ 1 + 1, . . . , λ 1 + λ 2 must fill the next row up, and so on, giving T = Y . Stanley symmetric functions. A reduced expression is a sequence ρ = (i k , . . . , i 1 ) such that the permutation s i k · · · s i 1 has k inversions, where s i is the simple transposition that interchanges i and i + 1. Define the set R(w) of reduced expressions for w by For example, the elements of R(42153) are shown in Figure 4. In order to enumerate reduced expressions, Stanley defined a family of symmetric functions indexed by permutations that are the generating functions for reduced expressions. Definition 2.5 ( [Sta84]). For w a permutation, the Stanley symmetric function S w is where the sum is over all reduced expressions for w. To avoid confusion with fundamental quasisymmetric functions, we diverge from usual notation of F w and denote the Stanley symmetric functions by S w . Also note that we follow usual conventions and have our S w = F w −1 in [Sta84]. We give an independent and elementary proof of this fact using dual equivalence. 2.3. Dual equivalence. Given a set of combinatorial objects A endowed with a notion of descents, one can form the quasisymmetric generating function for A by Two examples of this are Schur functions generated by standard Young tableaux (2.2) and Stanley symmetric functions generated by reduced expressions (2.4). Dual equivalence [Ass15] is a general framework for proving that such generating functions are symmetric and Schur positive. We recall the relevant definitions and theorems, then apply them to R(w) to prove that Stanley symmetric functions are symmetric and Schur positive. Definition 2.6 ([Ass15]). Let A be a finite set, and Des be a map from A to strong compositions of n. A dual equivalence for (A, Des) is a family of involutions {ϕ i } 1<i<n on A such that (i) For all i − h ≤ 3 and all T ∈ A, there exists a partition λ of i − h + 3 such that where [T ] (h,i) is the equivalence class generated by ϕ h , . . . , ϕ i , and Des (h,i) (T ) is the strong composition of i − h + 1 obtained by deleting the first h − 1 and last n − i parts from Des(T ). (ii) For all \|i − j\| ≥ 3 and all T ∈ A, we have For example, if Des(T ) = (3, 2, 3, 1), then we have Des (3,8) (T ) = (1, 2, 3). Haiman [Hai92] defined involutions d i on standard Young tableaux that swap i with i±1 whenever i ∓ 1 lies in between them in the column reading word (read top to bottom from left to right). Assaf [Ass15] showed that these involutions satisfy Definition 2.6 and that any involutions satisfying Definition 2.6 have Des-isomorphic equivalence classes. In particular, we have the following. for some partition λ. In particular, the fundamental quasisymmetric generating function for A is symmetric and Schur positive. We use the defining relations for the simple transpositions that generate the symmetric group to construct involutions on R(w) that satisfy Definition 2.6. For example, one of the two dual equivalences class of R(42153) is shown in Figure 6. Observe that the generating function for this class is the Schur function s (3,2) (X), picking off one term in the expansion S 42153 (X) = s (3,2) (X) + s (3,1,1) (X). (1, 2, 4, 1, 3) Remark 2.9. Note that if ρ contains distinct indices, then changing those indices, in order, to 1 . . . n and taking the inverse of the resulting permutation gives a bijection, say θ, with permutations, and we have θ(d i (ρ)) = d i (θ(ρ)), where d i is Haiman's dual equivalence involutions on permutations. Therefore, in this case, it follows from [Ass15] that the maps d i give a dual equivalence for R(w). Theorem 2.10. The maps {d i } give a dual equivalence for R(w). In particular, Stanley symmetric functions are symmetric and Schur positive. Proof. Both swaps, s i , and braids, b i , are themselves involutions on R(w) provided they apply only when the corresponding relation is valid for the simple transpositions s i , namely for s i whenever \|ρ i − ρ i+1 \| > 1 and for b i whenever ρ i−1 = ρ i+1 = ρ i ± 1. The definition of d i clearly ensures this. The conditions for when to apply a swap or a braid are maintained by the swap or braid, ensuring that d i is an involution. Note that d i changes ρ only at indices i − 1, i, i + 1, and the conditions that determine which swap or braid to apply look only at these indices. Therefore, if \|i − j\| ≥ 3, then {i − 1, i, i + 1} and {j − 1, j, j + 1} are disjoint, the maps d i and d j commute. It remains to consider restricted dual equivalence classes under d h , . . . , d i for i − h ≤ 3. For i − h = 0, d i acts trivially if and only if ρ i−1 ρ i ρ i+1 is weakly increasing or weakly decreasing. Since, by the reduced condition, consecutive letters may not be equal, the sequence must be strict, and so the descent composition is (3) (to get s (3) (X)) or (1, 1, 1) (to get s (1,1,1) (X)). If d i acts nontrivially, then it pairs acb with cab where a < b < c. In all cases, the two descent compositions are (2, 1) and (1, 2) giving s (2,1) (X). A similar by hand analysis can handle the cases i − h = 1, 2, 3, though it is more efficient (and perhaps less error-prone) to program the involutions on a computer and check them for all reduced words on {1, . . . , 2(i − h + 3)}, since we need only separate the cases \|a − b\| = 0, 1 or \|a − b\| > 1. A dual equivalence {ϕ i } for (A, Des) induces a Des-preserving map Φ from A to SYT such that, for any dual equivalence class C for A under {ϕ i }, the restriction of Φ to C gives a bijection with SYT(λ) for some (unique) partition λ. Definition 2.11. Given a dual equivalence {ϕ i } for (A, Des), we call the induced map Φ : A → SYT the rectification map for A with respect to {ϕ i }. For A ∈ A, we say that A rectifies to Φ(A). For example, for reduced expressions for 42153, the reduced expressions in Figure 6 rectify to the standard Young tableaux in Figure 5, respectively. Definition 2.12. Given a permutation w, say that a reduced expression ρ ∈ R(w) is super-standard if it rectifies to a super-standard tableau. For example, the super-standard reduced expressions for 42153 are shown in Figure 7. Corollary 2.13. For w a permutation, we have where c w,λ is the number of super-standard reduced expressions for w with descent composition λ. Key positivity of Schubert polynomials 3.1. Schubert polynomials. Lascoux and Schützenberger [LS82] defined polynomial representatives for the Schubert classes in the cohomology ring of the complete flag variety. The importance of these Schubert polynomials lies in the fact that their structure constants give intersection multiplicities for the corresponding varieties. Finding a combinatorial rule to compute these number remains one of the fundamental open problems in algebraic combinatorics. We refer the reader to [Mac91] for a beautiful and thorough treatment of the underlying combinatorics of Schubert polynomials, insofar as it is understood. As with the Schur case, we will harness the power of another basis, in this case the fundamental slide basis [AS17] of Assaf and Searles, to express Schubert polynomials as the generating function for reduced expressions. ). For a weak composition a of length n, define the fundamental slide polynomial F a = F a (x 1 , . . . , x n ) by where b ≥ a means b 1 + · · · + b k ≥ a 1 + · · · + a k for all k = 1, . . . , n. For example, we have Whereas fundamental quasisymmetric functions are indexed by strong compositions, fundamental slide polynomials are indexed by weak compositions, so we require a weak descent composition to define generating functions with respect to this basis. We adopt the following from [Ass17]. . Define the weak descent composition of ρ, denoted by des(ρ), by des(ρ) r i = \|ρ (i) \| and all other parts are zero if all r i > 0 and des(ρ) = ∅ otherwise. For example, among the reduced expressions in Figure 4, all but the first and fourth in the top row are virtual, and these have weak descent compositions (3, 1, 0, 1) and (3, 2, 0, 0), respectively. To facilitate virtual objects, we extend notation and set Building on the monomial model given by Billey, Jockusch, and Stanley [BJS93], Assaf [Ass17] gave the following expansion of Schubert polynomials in terms of fundamental slide polynomials, which we take as our definition. Definition 3.3 ([Ass17] ). For w any permutation, we have where the sum may be taken over non-virtual reduced expressions ρ. It is easy to see that v(λ, k) gives a bijection between grassmannian permutations with unique descent at k and partitions of length at most k. Moreover, we have the following. Theorem 3.4 ( [LS82]). For λ a partition and k a positive integer, we have Therefore the Schubert polynomials contain the Schur polynomials as a special case. However, we argue that Schubert polynomials more closely parallel Stanley symmetric functions than they do Schur functions, noting that the latter are also a special case of the former. Let 1 m × w denote the permutation obtained by adding m to all values of w in one-line notation and pre-pending 1, 2, . . . , m. Note that the reduced expressions for 1 m × w are simply those for w with each index increased by m. Let 0 m × a denote the weak composition obtained by pre-pending m zeros to a. Then for ρ ∈ R(w) non-virtual, the corresponding reduced expression for R(1 m × w) will have weak descent composition 0 m × des(ρ). To make our running example slightly more interesting, consider 1 × 42153 = 153264. From Figure 8, we have (5, 3, 2, 3, 4) (5, 2, 3, 2, 4) (5, 2, 3, 4, 2) (3, 5, 2, 3, 4) (3, 2, 5, 3, 4) (3, 2, 3, 5, 4) (2, 3, 5, 2, 4) Note that the fundamental slide expansion of S 1 m ×42153 gains no additional terms when m is at least two. Macdonald [Mac91] In parallel to this, Assaf and Searles [AS17] showed that fundamental quasisymmetric functions are the stable limits of fundamental slide polynomials. Therefore flattening the strong compositions in the fundamental slide expansion of S 1 m ×w precisely gives the fundamental quasisymmetric expansion of S w . To fit Schur functions into this stable picture, we consider another basis for the polynomial ring called key polynomials. Key polynomials. The key polynomials first arose as Demazure characters for the general linear group [Dem74] and were later studied combinatorially by Lascoux and Schützenberger [LS90] who expounded on their connection with Schubert polynomials. As with Schubert polynomials, original definitions were given in terms of divided differences, though we will derive a combinatorial model in terms of fundamental slide polynomials based on work of Kohnert [Koh91] and Assaf and Searles [AS16]. See [RS95] for a thorough treatment of the combinatorics of key polynomials. A diagram is a finite collection of cells in the Z × P lattice. We index each cell of a diagram by its top right corner. A diagram is virtual if it contains a cell with nonpositive row index. The weight of a diagram D ⊂ P × P, denoted by wt(D), is the weak composition whose ith part is the number of cells in row i of D. The weight of a virtual diagram is ∅. Definition 3.7 ([AS16]). Given a weak composition a of length n, a Kohnert tableau of shape a is a diagram filled with entries 1 a 1 , 2 a 2 , . . . , n an , one per cell, satisfying the following conditions: Figure 9. The key diagram for (3, 0, 4, 2, 3) (left) and another diagram with the same weight (right). (i) there is exactly one i in each column 1 through a i ; (ii) each entry in row i is at least i; (iii) the i's weakly descend from left to right; (iv) if i < j appear in a column with i above j, then there is an i right of and strictly above j. Kohnert tableaux are so named because they are based on Kohnert moves on key diagrams. The key diagram of a weak composition a, denoted by D(a), is the diagram with a i cells left-justified in row i. For example, the left diagram in Figure 9 is the key diagram for (3, 0, 4, 2, 3). A Kohnert move on a diagram selects a nonempty row and the rightmost cell therein, then pushes this cell down to the highest empty space below it. Kohnert [Koh91] showed that set of the diagrams obtained from Kohnert moves on a key diagram generate the key polynomial. Assaf and Searles [AS16] showed that these diagrams are in bijection with Kohnert tableaux, the latter being easier to enumerate directly. For example, Figure 10 gives QKT(0, 3, 0, 2). From this we compute κ (0,3,0,2) = F (0,3,0,2) + F (2,2,0,1) + F (1,3,0,1) + F (2,3,0,0) . We may reverse Kohnert moves on quasi-Yamanouchi Kohnert tableaux to give a simple tableau model for key polynomials in terms of certain fillings of key diagrams. Definition 3.10. A standard key tableau is a bijective filling of a key diagram with {1, 2, . . . , n} such that rows weakly decrease and if some entry i is above and in the same column as an entry k with i < k, then there is an entry immediately right of k, say j, and i < j. Definition 3.11. For a standard tableau T , the run decomposition of where τ is the decreasing word n · · · 21 broken between i + 1 and i precisely when i + 1 lies weakly right of i in T . In this case, we call i a descent of T . For example, the run decompositions for the standard key tableaux in Figure 11 are (54\|321), (5\|43\|21), (5\|432\|1), (54\|32\|1), (543\|21), respectively. Remark 3.13. Note that, for the current case of standard key tableaux, it is enough to take t i = min(row(τ (i) 1 ), t i+1 − 1). To see why, if k + 1, k are in τ (i) with k in row r strictly below, then since k is strictly right of k + 1, there must be some larger entry, say ℓ, left of k in its row and in the column of k + 1. In this case, ℓ must be in τ j for some j > i and we must have t j ≤ r. Therefore t i will not attain its value at r, the row of k. Despite this apparent simplification, we keep this more general definition for weak descent compositions as it is needed in § 4. For example, weak descent compositions for the standard key tableaux in Figure 11 are shown. Definition 3.14. For D ∈ QKT(a), the ascended tableau of D, denoted by A(D), is obtained by re-labeling the cells of D along rows from left to right, beginning at the top, with n, n − 1, . . . , 2, 1, and returning any cell originally labeled by i back to row i. For T ∈ SKT(a), the descended diagram of T , denoted by D(T ), is the diagram obtained by pushing cells down minimally until the word obtained by reading entries right to left, from bottom to top is the identity, and then relabeling cells based on their original row index. For example, the ascended tableaux for the quasi-Yamanouchi Kohnert tableaux in Figure 10 are given in Figure 11, respectively. Conversely, the descended diagrams for the standard key tableaux in Figure 11 are shown in Figure 10, respectively. Proof. Consider first A(D) for D a Kohnert tableau for a. Condition (i) of Definition 3.7 ensures that A(D) is a labeling of the key diagram of a; condition (ii) ensures that cells move weakly up in passing from D to A(D); condition (iii) ensures that rows of A(D) are weakly decreasing; and condition (iv) ensures that if some entry i of A(D) lies above some entry k with i < k in the same column, then there is any entry j immediately right of k with i < j. In particular, A(D) ∈ SKT(a). Next consider D(T ) for T ∈ SKT(a). Re-labeling cells based on original row index ensures condition (i) of Definition 3.7 holds for D(T ); cells moving down ensures condition (ii); entries weakly decreasing along rows ensures that cells to the right in the same row move weakly lower, giving condition (iii); and the column inversion condition for key tableaux precisely corresponds to condition (iv). Therefore D(T ) is a Kohnert tableau for a. Moreover, for every nonempty row r of the key diagram for a, either the leftmost entry, say i, remains in row r or it must be pushed minimally down, in which case sits in the row immediately below i + 1 which, by the definition of descents, lies weakly to its right. Thus D(T ) is quasi-Yamanouchi. With images established, the maps are clearly inverse to one another, proving that both are indeed bijections. For D ∈ QKT(a), after we re-label cells, i will be a descent of A(D) if and only if i is the leftmost in its row. In particular, wt(D) = des(A(D)). In particular, standard key tableaux give another characterization of key polynomials. Corollary 3.16. The key polynomial for a weak composition a is given by where the sum may be taken over non-virtual standard key tableaux of shape a. Given a partition λ and a positive integer k that is at least the length of λ, let a(λ, k) denote the weak composition of length k with weakly increasing parts that sort to λ. Then we have the following. Therefore the key polynomials also contain the Schur polynomials as a special case. We argue that the parallel here is much deeper than with Schubert polynomials. We say that a standard key tableau is yamanouchi if its reverse row reading word is the identity. We have the following key tableau analog of Proposition 2.3. Proof. Clearly Y is unique, if it exists, and it can be constructed by placing entries 1, . . . , a 1 left to right in row 1, a 1 + 1, . . . , a 1 + a 2 left to right in row 2, and so on. Then i will be a descent of Y precisely for the partial sums i > 0 in {a 1 , a 1 + a 2 , . . .}, giving des(Y ) = a. Conversely, if Des(T ) = a, then the upper uni-triangularity of key polynomials with respect to monomials evident from Kohnert's expansion shows that D(T ) is the key diagram for a, in which case T = Y . Standard key tableaux provide the natural analog for standard Young tableaux in our generalization of dual equivalence. Comparing key tableaux for a with those for 0 m × a, prepending 0's simply prepends 0's to the weak descent composition. Note that the number of terms in the fundamental slide expansion of κ 0 m ×(3,2) remains the same when m is at least two. This stability phenomenon is explained by the fact that Schur functions are the stable limits of key polynomials. This is implicit in work of Lascoux and Schützenberger and is made explicit in [AS16]. Furthermore, Assaf and Searles [AS16] proved that flattening the compositions in the fundamental slide expansion of κ 0 m ×a precisely gives the fundamental quasisymmetric expansion of s sort(a) . For example, flatten κ 0 2 ×(3,2) and compare with s (3,2) . As with Schur functions and Stanley symmetric functions, we give an independent and elementary proof of this by lifting dual equivalence to polynomials. 3.3. Weak dual equivalence. Given a set of combinatorial objects A endowed with a notion of weak descents, one can form the fundamental slide generating polynomial for A by Two examples of this are Schubert polynomials generated by reduced expressions (3.3) and key polynomials generated by standard key tableaux (3.9). We generalize the notion of dual equivalence to polynomials defined in this way as follows. Definition 3.20. Let A be a finite set, and let des be a map from A to weak compositions of n. A weak dual equivalence for (A, des) is a family of involutions {ψ i } 1<i<n on A such that (i) For all i − h ≤ 3 and all T ∈ A, there exists a weak composition a of i − h + 3 such that For example, if des(T ) = (0, 3, 2, 0, 3, 1), then des (3,8) (T ) = (0, 1, 2, 0, 3, 0). As a first example of weak dual equivalence, we construct a weak dual equivalence for standard key tableaux. Define the column reading order of a standard key tableau to begin at the lowest cell of the leftmost column, read entries in the column bottom to top, then continue with the next column to the right. For example, the column reading order for the leftmost tableau in Figure 11 is 35241. Definition 3.21. Given T ∈ SKT(a) and 1 < i < \|a\|, define d i (T ) as follows. Let b, c, d be the cells with entries i − 1, i, i + 1 taken in column reading order. Then where b j cycles j − 1, j, j + 1 so that j shares a row with j ± 1 and s j interchanges j and j + 1. Proof. Let T ∈ SKT(a) and, with notation as in Definition 3.21, suppose b, d are in the same row and c is not. We use the key tableaux condition that if i < k are in a column with i above k, then there is a j > i immediately right of k to argue that we have the case depicted in the left of Figure 12. First we claim that c does not have value i. If it did, then b must have value i + 1 and d value i − 1, in which case they must be adjacent in their row. Since i lies between in column reading order, it must lie above i + 1, a contraction of the key tableaux condition since i < i + 1 but i > i − 1, or below i − 1, a contradiction again since anything to the right of i must be smaller than i − 1. Therefore c is either i − 1, in which case it cannot be below i since the entry to its left will be below and larger than i + 1, or c is i + 1, in which case it cannot sit below i − 1 since any entry to its right will be below and smaller than i − 1. Thus we have c above b and d i (T ) = b i (T ) acts as shown in Figure 12. Next suppose b and d lie in different rows. The same key tableaux condition ensures that if c does not have entry i, then b and d must also lie in different columns. In this case, all entries in the same row or column as b or d compare the same with i and i ± 1, so the key tableaux conditions are preserved by whichever of s i−1 or s i applies. Again, notice the reversal of lengths. Note also that Des(T ) = flat(des(T )). It is clear that inserting or deleting rows with no cells does not change the allowable fillings of the standard key tableaux. That is, there is an obvious Des-preserving bijection between SKT(a) and SKT(b) whenever flat(a) = flat(b). The Des-preserving bijection between SKT(a) and SKT(b) whenever sort(a) = sort(b) is less obvious. It follows as a corollary to the following. Proof. Consider the map Φ on SKT(a) defined by letting the cells of T fall to shape sort(a), then sorting the columns to descend upward, and replacing i with n − i + 1. After letting entries fall and sorting columns, the rows will necessarily be decreasing left to right, and so Φ(T ) ∈ SYT(sort(a)). Note that i + 1 lies strictly right of i in T if and only if n − i lies strictly right of n − i + 1 in Φ(T ). In particular, i ∈ Des(T ) if and only if n − i ∈ Des(Φ(T )). Moreover, we claim that Φ(d i (T )) = d n−i+1 (Φ(T )). To see this, note that, by the ever useful key tableaux condition, if i and i + 1 appear in the same column of T , i + 1 must be above i since anything to the right of i + 1 is smaller than i. Therefore the column reading word for T restricted to i − 1, i, i + 1 maps to that for Φ(T ) restricted to n − i, n − i + 1, n − i + 2. Then s i−1 and s i correspond under Φ to exchanging the larger two and smaller two entries, respectively, and, upon sorting columns, b i corresponds to swapping the smaller two, thus establishing the claim. The theorem now follows from the symmetry of Schur functions, and the corollary from the super-standard characterization of tableaux. i is a descent of T , so the restricted weak descent composition flattens to either (1, 1, 1) or (3). In either case, the corresponding key polynomial is a single fundamental slide polynomial, and so the equivalence class corresponds to a single key polynomial. If d i (T ) = b i (T ), then we may assume T has c = i + 1. Then the restricted run decomposition of T is i + 1\|ii − 1, and that of b i (T ) is i + 1i\|i − 1. Therefore des (i−1,i+1) (T ) = (0 m , 2, 0 n , 1), where n = 0 if and only if either i + 1 lies in the row immediately above i and i − 1 or if t j of the block containing i + 1 is forced to be t j+1 − 1. Either way, we have des (i−1,i+1) (b i (T )) = (0 m−1 , 1, 2), and so the equivalence class corresponds to the polynomial F (0 m ,2,0 n ,1) + F (0 m−1 ,1,2) = κ (0 m ,2,0 n ,1) . If d i (T ) = s i−1 (T ), and so, again, c = i + 1. We may assume T has i − 1 left of i. Then the restricted run decomposition of T is i+1i\|i−1, and that of s i−1 (T ) is i+1\|ii−1. Moreover, in this case, for both T and s i−1 (T ), both blocks of the run decomposition must be forced to have t j = t j+1 − 1 since there is an entry larger than i left of i and below i − 1. Therefore des (i−1,i+1) (T ) = (0 m , 1, 2) and des (i−1,i+1) (s i−1 (T )) = (0 m , 2, 1), and so the equivalence class corresponds to the polynomial F (0 m ,1,2) + F (0 m ,2,1) = κ (0 m ,1,2) . The argument for d i (T ) = s i (T ) is completely analogous. Again, one can either carry out similar analyses for the cases i − h = 1, 2, 3, or, to avoid tedium, since the action of d i is determined by relative positions of these cells based on their rows and columns, there are finitely many configurations to check by computer. Under certain stability assumptions, the converse of Theorem 3.25 also holds. Definition 3.26. The key polynomial κ a is F-stable if both κ a and κ 0 m ×a have the same number of terms in their F-expansions for any m > 0. To make this condition easier to establish, we have the following proposition proved in [AS16]. Proposition 3.27. The following conditions are equivalent: (1) SKT(a) contains no virtual elements; (2) the number of terms in the F-expansion of κ a is the size of SYT(sort(a)); (3) both κ a and κ 0 m ×a have the same number of terms in their F-expansions for some m > 0; (4) both κ a and κ 0 m ×a have the same number of terms in their F-expansions for any m > 0. In particular, κ a is F-stable if and only if any one of these conditions is met. Definition 3.28. A weak dual equivalence for (A, des) is stable if the restricted dual equivalence classes of degrees up to 6 are F-stable key polynomials. While Definition 3.28 looks like a local condition, the following result shows that it is global. Theorem 3.29. Let A be a set of combinatorial objects for which des is never ∅ (i.e. A has no virtual elements). If {ψ i } is a stable weak dual equivalence for (A, des), and U ∈ A, then (3.14) F des(T ) = κ a for some key-stable weak composition a. In particular, the fundamental slide generating polynomial for A is key positive. Proof. We may assume A has a unique equivalence class under {ψ i }. Since each κ a appearing as the generating polynomial for a restricted equivalence class is stable, there is a Des-preserving bijection SKT(a) → SYT(sort(a)). Therefore ψ induces a dual equivalence on (A, Des). In particular, since there is a unique equivalence class, we have a Des-preserving bijection Φ : A ∼ → SYT(λ) for some partition λ. Given any weak composition a for which sort(a) = λ, by Theorem 3.24, we may extend Φ to a Des-preserving bijection Ψ a : A ∼ → SKT(a). We will show by induction on the size of λ that there exists a (unique) weak composition a with sort(a) = λ such that Ψ a is des-preserving. If λ is a single row, then A has a single element, say A, and we may take a = des(A). Assume, then, that λ has at least two rows, as we proceed by induction on \|λ\|. For \|λ\| ≤ 6, this follows immediately from the definition, thus establishing the base case. Assume \|λ\| = n > 6, and assume the result for strictly smaller partitions. Let m = #{λ i } be the number of removable corners of λ. Break A into equivalence classes under ψ 3 , . . . , ψ n−1 , say A (1) , . . . , A (m) . By induction, for each i there is a bijection Ψ (i) : A (i) ∼ → SKT(â (i) ) for some unique weak compositionâ (i) of n − 1 such that des (2,n) (A) = des(Ψ (i) (A)). For fixed i, let m i ≥ 0 be the length of the row of sort(â (i) ) to which a cell is added to obtain λ (since we necessarily have sort(â (i) ) ⊂ λ). Let k i be the lowest part ofâ (i) that equals m i if m i > 0, or take k i to be the largest index that occurs as the smallest nonzero entry among des(A) for A ∈ A (i) . Set a (i) to be the weak composition of n obtained by adding 1 toâ (i) k i . Then we can lift Ψ (i) to a des-preserving injection Ψ (i) : A (i) ∼ → SKT(a (i) ) where the image is those key tableaux with 1 in row k i . We claim that a (i) = a (j) = a, in which case the injections Ψ (i) combine to give the desired des-preserving bijection Ψ : A ∼ → SKT(a). Using the injections Ψ (i) , we may identify each A ∈ A (i) with an element of SKT(a (i) ). We recall some basic properties inherited from dual equivalence. For any A ∈ A, we may use ψ 5 , . . . , ψ n−1 to move 4, . . . , n into any positions which they can occupy with 1, 2, 3 in some fixed positions. Moreover, ψ j ψ 2 = ψ 2 ψ j for any j ≥ 5 and des (4,n) (ψ 2 (A)) = des (4,n) (A). Given i, j, there exists B ∈ A (i) and C ∈ A (j) such that C = ψ 2 (B). Combining this, we must a des (4,n) -preseving bijection between the class of B generated by ψ 5 , . . . , ψ n−1 . In particular, the shapes of a (i) and a (j) must agree after deleting the fixed positions for 1, 2, 3 from B and C, respectively. Therefore the result follows from the fact that the weak compositions under ψ 2 , ψ 3 , ψ 4 must give a key polynomial. As demonstrated in the proof of Theorem 3.29, a stable weak dual equivalence {ψ i } for (A, des) induces a des-preserving map Ψ from A to SKT such that, for any dual equivalence class C for A under {ψ i }, the restriction of Ψ to C gives a bijection with SKT(a) for some (unique) weak composition a. Definition 3.30. Given a stable weak dual equivalence {ψ i } for (A, des), we call the induced map Ψ : A → SKT the weak rectification map for A with respect to {ψ i }. For A ∈ A, we say that A weakly rectifies to Ψ(A). (6, 4, 3, 4, 5) Lemma 3.31. Let w be a permutation for which no element of R(w) is virtual. Then the maps d i on reduced expressions give a stable weak dual equivalence for (R(w), des). Proof. By Theorem 2.10, the maps {d i } are involutions on R(w), and d i and d j commute for \|i − j\| ≥ 3. Therefore we need only consider restricted dual equivalence classes under d h , . . . , d i for i − h ≤ 3. We begin with the analysis of descent compositions in the proof of Theorem 2.10 and consider the consequences for considering weak descent compositions instead. For i − h = 0, d i acts trivially if and only if ρ i−1 ρ i ρ i+1 is weakly increasing or weakly decreasing. Since, by the reduced condition, consecutive letters may not be equal, the sequence must be strict, and so the descent composition is (3) or (1, 1, 1). Since κ a = F a for a any weak composition that flattens to (n) or (1 n ), in either case the class is a single key polynomial. As before, a similar by hand analysis can handle the cases i − h = 1, 2, 3, though it is more efficient (and certainly less error-prone) to program the involutions on a computer and check them for all reduced words on {1, . . . , i − h + 3}. In particular, we have a simple, combinatorial proof of the key positivity of Schubert polynomials. Note that, as we show below, this holds for any permutation w, not only for those with no virtual reduced expressions. For example, for reduced expressions for 1 2 × 42153, the reduced expressions in Figure 14 rectify to the standard key tableaux in Figure 15, respectively, after stabilizing by prepending 0 2 . We may equally well use the induced bijection between the corresponding reduced expressions for 42153 and SKT (3, 1, 0, 1), where the virtual terms coincide. Definition 3.32. Given a permutation w, say that a reduced expression ρ ∈ R(w) is yamanouchi if it weakly rectifies to a yamanouchi key tableau. For example, the yamanouchi reduced expressions for 42153 are shown in Figure 16. Theorem 3.33. For w a permutation, we have where c w,a is the number of yamanouchi reduced expressions for w with weak descent composition a. In particular, the Schubert polynomial S w is key positive. Proof. For S w an F-stable polynomial, the result follows from Lemma 3.31 and Theorem 3.29. For w not F-stable, by [AS17] there exists a (specific) nonnegative integer η(w) for which 1 η(w) × w is F-stable. Let Ψ be the induced des-preserving map from R(1 η(w) × w) to SKT. For any nonvirtual elements of R(w), the corresponding reduced expressions for R(1 η(w) × w) are precisely those that map to some standard key tableau with weak descent composition 0 η(w) × a. Given any weak dual equivalence class, we may pull back both the reduced expressions in R(1 η(w) × w) that are nonvirtual in R(w) and those standard key tableaux that have at least η(w) leading 0's. This gives a des-preserving bijection between nonvirtual elements, so they must have the same generating polynomial. Hence S w is also key positive with the same leading terms. Littlewood-Richardson rules 4.1. Shuffle products and Schur products. Gessel used the shuffle product of Eilenberg and Mac Lane [EML53] to give a Littlewood-Richardson rule for fundamental quasisymmetric functions [Ges84]. The shuffle product of words A and B, denoted by A ¡ B, is the set of all ways of riffle shuffling the terms of A, in order, with the terms of B, in order. denotes the multiplicity of γ in the shuffle product α ¡ β. Given partitions µ, ν, define the diagram µ ⊗ ν to be the concatentation of Young diagrams for µ and ν. A standard Young tableau of shape µ⊗ν is a bijective filling of µ⊗ν with entries {1, 2, . . . , n} such that each shape satisfies the Young tableaux conditions. Extend descent compositions to product shapes using the column words with entries of µ read before entries of ν. For example, the standard Young tableau of shape (3, 2)⊗(2, 1) on the right side of Figure 17 has descent composition (3, 2, 3). The following expansion is an immediate consequence of Theorem 4.2. Proposition 4.3. For partitions µ, ν, we have We may extend the dual equivalence operators d i to products of tableaux using the column reading word to obtain the classical Littlewood-Richardson rule for Schur functions. where c λ µ,ν is the number of super-standard tableaux of shape µ ⊗ ν that rectify to λ. Remark 4.5. Dual equivalence can rediscover the jeu de taquin algorithm of Schützenberger [Sch77] that gives an explicit rectification process. Indeed, Haiman originally called his involutions dual equivalence since they are precisely dual to jeu de taquin. That is, dual equivalence moves commute with jeu de taquin moves and as such can be used to give a simple proof that jeu de taquin is well-defined and provides the explicit map from skew shapes to straight shapes needed to prove Theorem 4.4. Adjoint to products of shapes, whenever µ ⊆ λ we may define the skew shape λ/µ to be the set-theoretic difference between the two shapes. Define the corresponding skew Schur function by (4.5) Once again, the dual equivalence operators on tableaux apply, giving the following. Since skewing is adjoint to multiplication, the repetition of notation in Theorems 4.4 and 4.6 is intentional. That is, c λ µ,ν is well-defined by either and agrees for both. 4.2. Skew key polynomials. Given weak compositions a, d, we say that a ⊆ d if a i ≤ d i for all i. Equivalently, the key diagram for a is a subset of the key diagram for d. This allows us to form the skew key diagram d/a as the set-theoretic difference between the two key diagrams. We extend the notion of standard key tableaux to skew shapes, but now we allow a new type of column inversion. Definition 4.7. For weak compositions a ⊆ d, a standard skew key tableau of shape d/a is a bijective filling of the skew key diagram d/a with entries {1, 2, . . . , n} such that rows weakly decrease and if some entry i is above and in the same column as an entry k with i < k, then either there is an entry immediately right of k, say j, and i < j or there is a skewed cell immediately left of k and an entry j < k immediately left of i. Theorem 4.9. The maps {d i } give a dual equivalence for (SKT(d/a), Des). In particular, where c λ µ,ν is the number of super-standard skew tableaux of shape d/a that rectify to ν. Proof. It does not follow from Lemma 3.22 that d i is well-defined on skew diagrams. Certainly if d i acts by swapping i and i ± 1, the two entries are not in the same row or column, so the map is well-defined in this case. However, for b i , we must be more careful since now we can have two of i, i ± 1 in the same row with i ∓ 1 in a lower row if it lies in the right column with a skewed cell to its left (for example, see the second and third tableaux from the left in Figure 19). This case is effectively a rotation of the braid case for straight diagrams, so this action is also well-defined. Consider the bijection Φ : SKT(a) ∼ → SKT(sort(a)) from the proof of Theorem 3.24 defined by letting the cells of T fall to shape sort(a), then sorting the columns to descend upward, and replacing i with n − i + 1. If we have skew cells regarded as smaller than any others, then this maps extends to the skew case and still commutes with the maps d i in the sense that Φ(d i (T )) = d n−i+1 (Φ(T )), hence it is a bijection Φ : SKT(d/a) ∼ → SKT(sort(d)/sort(a)) and the result follows. As with the case of products, these maps do not, in general, give a weak dual equivalence as evidenced by Figure 19. However, as with products, they do in the case when a is nondecreasing. Theorem 4.10. For λ a partition of length n, let a λ be the weakly increasing weak composition of length n that sorts to λ. For d any weak composition of length n with a ⊆ d, the maps {d i } give a weak dual equivalence for (SKT(d/a λ ), des). In particular, whereĉ d a λ ,b is the number yamanouchi skew key tableaux of shape d/a λ that rectify to b. Proof. When a λ is a partition shape that sits at the top row of d, then if i lies strictly below and strictly right of i + 1, then there must be a cell (not skewed) containing an entry, say k, below i + 1 in the same column and left of i in the same row, and necessarily k > i + 1 so k lies in a different block of the run decomposition than i + 1 and i. Therefore in constructing the weak descent composition, the run block containing k will be indexed by some row weakly below that of i, and so the run block for i + 1 will be indexed by some row strictly below i. Thus we may equivalently define the weak descent compositions by taking the largest entry, instead of the lowest entry, and now the definition coincides with that for a non-skewed key tableaux. Therefore the result follows from Theorem 3.25. For example, each row of standard skew key tableaux in Figure 20 is a dual equivalence class, and from the weak descent compositions we compute κ (3,2,3)/(0,1,2) = κ (3,1,1) + κ (3,2,0) . Remark 4.11. We can describe the rectification rule more directly by first applying the injection SKT(d/a) → SYT(sort(d)/sort(a)), then rectifying according to jeu de taquin, then applying the bijection SYT(µ) ∼ → SKT(c). In so doing, one sees that the rectification map can be described more directly by a process similar to jeu de taquin, though now entries may cycle when a cell slides left, so the explicit algorithm becomes more involved. 4.3. Products of key polynomials. Assaf and Searles generalized the shuffle product to weak compositions to give a Littlewood-Richardson rule for fundamental slide polynomials [AS17]. Given weak compositions a, b, define the diagram D(a⊗b) to be the concatenation of key diagrams for a and for b. A standard key tableau of shape a ⊗ b is a bijective filling of D(a ⊗ b) with entries {1, 2, . . . , n} such that each diagram satisfies the key tableaux conditions. We now use the full power of Definition 3.12 to define weak descent composition for a product shapes. For example, the standard key tableau of shape (0, 2, 0, 3) ⊗ (0, 0, 2, 1) on the right side of Figure 21 has run decomposition (876\|54\|321) and weak descent composition (3, 2, 3, 0). to SKT((2, 0, 3) ⊗ (0, 2, 1)). Theorem 4.14. For weak compositions a, b, we have Extend dual equivalence operators d i from Definition 3.21 to standard key tableaux on products by asserting in Definition 3.21 that d i (T ) = T whenever c = i (this is implicit in the original definition as shown in the proof of Lemma 3.22). For example, Figure 22 gives one dual equivalence class for SKT(0, 2, 1, 0) × SKT(0, 1, 0, 1). Note that the generating function is s (3,2) and the generating polynomial is κ (0,3,2,0) . where c λ µ,ν is the number of super-standard tableaux of shape µ ⊗ ν that rectify to λ. Proof. By Proposition 4.15, the maps {d i } are involutions, and d i and d j commute for \|i − j\| ≥ 3. Therefore we need only consider weak descent compositions of restricted dual equivalence classes, and for this we must consider how the weak descent composition can differ from the case of a single key tableau. Note that the key tableaux conditions ensure that SKT a λ has decreasing rows (left to right) and columns (top to bottom). The only place for discrepancy with Theorem 3.25 is if some entry, say i, in a λ lies in the same run block but in a lower row than an entry, say j > i, in b. However, in this case, there is necessarily an entry, say k, immediately above i and k > j since k > i and not in the same run block. Therefore, in constructing the weak descent composition, the run block containing k will be indexed by the row above i, or lower, and so the run block for j will be indexed by the row of i, or lower. Therefore we may equivalently define the weak descent compositions by taking the largest entry, instead of the lowest entry, and now the definition coincides with that for a single key tableaux. Therefore the result follows from Theorem 3.25. Notice that the proof of Theorem 4.16 is dependent on placing the partition key diagram to the right of the arbitrary key diagram. While the generating polynomials clearly commute, the obvious bijection between SKT(a λ ⊗ b) and SKT(b ⊗ a λ ) that swaps the two tableaux is not des-preserving. Since Schubert polynomials are polynomial representatives for Schubert classes in the cohomology ring, the structure constants for Schubert polynomials, c w u,v , defined by enumerate flags in a suitable triple intersection of Schubert varieties. Therefore these so-called Littlewood-Richardson coefficients are known to be nonnegative. A fundamental open problem in Schubert calculus is to find a positive combinatorial construction for c w u,v . Recalling that in the special case of the Grassmannian subvariety, Schubert polynomials are Schur polynomials [LS82] which, in turn, are key polynomials, we have the following geometrically significant corollary to Theorem 4.16. where a = a(µ, m), b = a(ν, n) and c d a,b is the number of Yamanochi key tableaux of shape a ⊗ b that rectify to d. That is, we have a combinatorial rule for the key expansion of any arbitrary product of grassmannian Schubert polynomials. Note that the terms on the right side are not, in general, Schur polynomials. For example, s (1,1) (x 1 , x 2 , x 3 )s (1) (x 1 , x 2 ) = κ (1,1,1) + κ (0,2,1) , and the latter term on the right is not symmetric.	2017-02-14T02:42:21.000Z	2017-02-14T00:00:00.000	{ "year": 2017, "sha1": "521e155378ff3ce7f17ed3b0be6f793205054a0d", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "521e155378ff3ce7f17ed3b0be6f793205054a0d", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics" ] }
219650131	pes2o/s2orc	v3-fos-license	UNMET NEED OF HEALTH FACILITIES IN BOGOR REGENCY The existence of health facilities (fasilitas kesehatan/faskes) especially in district levels is very suitable to reduce the number of people with impaired level of health. For this reason, a targeted program which aims to reduce the number of unmet needs in areas that have high population, high rate of increase and high density is badly needed. This is in line with the characteristics of Bogor regency as a research locus. The method used in this research was a combination method between tabular analysis (statistics) and spatial data. The combination combined locations with low unmet need health facilities resulting from a causality of 2 (two) variables, the number of residents and the number of health facilities, so that the precise and accurate results could be obtained. The conclusion from the spatial analysis is that 19 out of 40 sub-districts in Bogor regency were indicated unmet need for low health facilities. Although this research was very simple, the result of this study is very important because it can be used as a guideline for the development of health facilities, especially community health centers (Puskesmas) in Bogor regency. Introduction The development of health facilities is intended to provide services for residents with impaired health (Badan Pusat Statistik (BPS)/ the Central Bureau of Statistics, 2018). However, the existence of these facilities is not yet made use of by residents for various reasons, such as the absences of costs for treatment and transportation, the long waiting -time or the unavailability of the nearest health facilities that can be reached so that it is hard for treatment. The indicator to measure the number of people who are sick but not treated because of those above factors is called unmet need for health services (Badan Pusat Statistik (BPS)/ the Central Bureau of Statistics, 2018). One step to overcome this problem appropriately and quickly is to increase the development of health facilities, especially for administrative areas that are close to resident population i.e. districts (Ministry of Health, 2015). According to the Ministry of Health, (2015), information on this image in the past three years has decreased in line with high population growth. Bogor Regency was chosen as the location of the study because this region has a unique characteristic that is this area has a very wide area, high population growth rate, high population density and high disrupted population (BPS Bogor Regency, 2018). However, the number of health facilities is still limited and not evenly spread in all districts. Researches related to unmet need for health facilities were already conducted by several researchers, for example (Septarini, 2017), (Ariyanti, 2017), (Masseria, 2009), (Hwang, 2018) and (OJ, 2016). Unlike those researches, this study used different and new method to be used in determining the value of unmet need facilities, especially Puskesmas (public health center) at district level, which was a combined method using tabular data (statistics) and spatial data in the form of spatial distribution geo-tagging points (spatial). Method Determination of the location of the unmet need can be declared successful and accurate if the determination of parameters/ variables and methods used are accurate and precise. There are 3 (three) parameters used and all of them are generated from the census and survey activities carried out by the Central Bureau of Statistics (BPS). Those three parameters are population data (BPS Bogor Regency, 2018), data on the number cases of health problems (BPS Bogor Regency, 2018) and data on geo-tagging points of the health infrastructure obtained from the WILKERSTAT application that were carried out in range of time of 2018-2019 There are 2 (two) stages of the method used in this study in order to produce precise and accurate results. In the outline, this study used a combination of methods whereas in an analysis, this study used 2 analysis: statistical and spatial. The first stage is the statistical analysis test phase. According to researchers, the statistical tests needed to be done are regression test and correlation between parameters. It is necessary to find out which parameters have high interdependence values. Regression analysis method is needed to find out the relationship between two variables (parameters), while the correlation analysis method is used to see patterns in one variable (parameter) based on the other parameters (Arikawati, 2016). So from both analysis, we will get a model that has high causal relationship (Arikawati, 2016). The second stage is spatial analysis. This is a newly used method, namely the method of "exploratory regression". To get the value of regression and spatial autocorrelation, we use the Moran's method all of which are found in the tool in ArcGIS 10.6 software. In this method, the regression value is seen from AICc (Akaike Information Criterion) value and from AdjR2 value. Based on research from Fathurahman, (2009), the value of AICc is a part of OLS (Ordinary Least Squares) method used to determine the best regression value, especially to forecast purposes (forecasting or referencing). Recommended value is the smallest value of all models used (Sinharay, 2010). While AdjR2 (R squared adjusted) value is used to see a regression pattern, whether increasing number of samples will increase the regression value (Hession and Moore, 2011). If the adjR2 value approaches the value of 1 along with addition of samples numbers, then the model is said to be the best model (Frost, 2019). The next thing to do in this study is to determine the value of spatial autocorrelations (SA) which is obtained using Moran's method. Wang (Wang, 2019) in his research said that this method is used to find two or more spatial data that are close together and have similarities. The next step is to determine the criteria for unmet need for a location. The determination of ideal ratio data for each category of health facility infrastructure is needed. This determination is in accordance with Septarini's thesis, (2017) saying that the ideal health ratio is 1 Puskesmas for 30,000 population. By using spatial analysis, "select by attributes", the unmet need location will be obtained by conducting a query process using population data and data on the number of geo-tagging points of Puskesmas infrastructure from the WILKERSTAT application. This data collection was conducted in 2018-2019. Result and Discussion The results obtained from determining the regression and correlation values can be seen in Table 1. From the two tables above, it can be seen that the influence of the population has a great impact to the number of health infrastructure points. In other words, the more population existed, the more health facilities must be available After obtaining the used parameters, the next step is to determine the location of unmet need. From this spatial method we get results as seen in image 1 below. Image 1. Unmet need location distribution map for PUSKESMAS infrastructure (Source: Data Processing) Based on image 1 above, the red colors on the map are 19 out of 40 districts in Bogor Regency that need health facilities i.e. Puskesmas, both for new infrastructure and infrastructure additions. The 19 districts can be seen in table 3 below. These findings are very precise and accurate because this study included more detailed elements of accuracy in taking geo-tagging in its application. This is in accordance with an article from Miller, (2013). This application for geo-tagging is called the WILKERSTAT application, which functions to take and photograph all infra points throughout Indonesia up to the RT/neighborhood/hamlet level. This application is more accurate and precise because of additional information from the chiefs/leaders of SLS/environmental/hamlet regarding information on the presence of health facilities in their areas. The difference of this research from the previous ones is that the results obtained in this study are aimed at calculating the lack of Puskesmas health facilities in each district in Bogor regency, whereas in previous studies the results are only aimed at calculating the ratio (Septarini, 2017). Although the results of this analysis are only focused on health facilities for public health center (Puskesmas) and only at district levels, when viewed from the map, the distribution of population is mixed with the data on health infrastructure points (Puskesmas) at regency level, the same pattern can be seen, which is only centered in the region close to the Bogor city and Depok city. Information is spread as in Image 2 below. In other words, the development is only focused in areas close to the city (Nugraha, 2012 This can also be seen in Image 3 below which shows a map of population distribution combined with data points of health infrastructure (hospitals) in one regency. Image 3. Map of population distribution with hospital infra-health availability (source: data processing) Conclusion and Suggestion A number of health facilities are needed especially in Bogor regency. This is because this region has a large population, the highest number of districts, the most extensive area and a high rate of population growth in the West Java province. Therefore, information about unmet need for health facilities is very much needed in the determination of policies, especially in terms of overcoming the number of health cases in this regency. The conclusions of this study based on the above analysis are as follows: a. The combination of statistical and spatial analytical methods is very helpful in determining the precise and accurate number of unmet need. b. Regression and correlation values are needed for the initial analysis test as a guide to which variables/parameters are interconnected and influential, where these variables/parameters will be used in spatial analysis. c. There are 19 out of 40 districts in the study location where the unmet need of health facilities numbers are still low. d. The pattern of distribution of health facilities is only centered in areas close to urban areas. This is very burdensome for people who live far from urban areas who want to seek treatment. e. There is a great need of additional health facilities and new health facilities development, in this case is Puskesmas, in several districts that have less unmet need numbers. The suggestion: a. There is a need of developing larger and more complete health facilities in terms of the completeness of medicines, such as hospitals, in several regions based on the ideal ratio for hospitals, which is 1 hospital is for 100.000 population, and also a need of medical personnel b. Provision of data on the number of health facilities must always be updated based on conditions in the field. c. The process of taking infra-health points (geo-tagging) must always be updated on an ongoing basis. d. The use of tools or gadgets for infrared point recording should use a detailed level of accuracy (<5 meters). e. For further research in the framework of infrastructure development or construction, new analytical methods are needed for the development of new health facilities that are friendly to community by using the "find near location" or "network analyst" method by looking at the elements of nearest mileage and access road that can be traversed. Acknowledgements I would like to thank to the Bogor Regency BPS for providing data on population and number of health cases, the Mapping Team in PPS Subdit of the Central Bureau of Statistics (BPS) who amazingly produced a spatial-based application for mapping boundaries and took points and photos for all infrastructure categories throughout Indonesia. Last but not least, special thank also goes to the teaching team of KTI (Karya Tulis Ilmiah/ Scientific Writing) which have always given encouragement to complete this research.	2020-05-21T00:12:17.835Z	2020-05-01T00:00:00.000	{ "year": 2020, "sha1": "9d3f023c80f781ab2d32882adf825780a86bc448", "oa_license": "CCBYSA", "oa_url": "https://ejournal.poltekkes-smg.ac.id/ojs/index.php/jrk/article/download/5387/1600", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "768dc5662fbc053de6c6c0a9d954aa7e3443cf5c", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Business" ] }
256039213	pes2o/s2orc	v3-fos-license	Amplitudes involving massive states using pure spinor formalism Same amplitudes evaluated independently using RNS and pure spinor formalism are expected to agree. While for massless states, this fact has been firmly established, for massive states such an explicit check has been lacking so far. We compute all massless-massless-massive 3-point functions in open supertrings in pure spinor formalism for the first massive states and compare them with the corresponding RNS results. We fix the normalization of the vertex operators of the massive states by comparing same set of 3-point functions for a fixed ordering in the two formalisms. Once fixed, the subsequent 3-point functions for each inequivalent ordering match exactly. This extends the explicit demonstration of equivalence of pure spinor and RNS formalism from massless states to first massive states. Introduction Since the turn of last century, the pure spinor formalism has emerged as a powerful alternative to the RNS and Green Schwarz formalisms for superstring theory [1][2][3]. The main advantage of the pure spinor formalism over RNS is that it maintains manifest spacetime super-Poincaré covariance at all stages. There are several general arguments demonstrating how the cohomology of the pure spinor formalism is equivalent to the cohomology of the RNS and Green Schwarz formalisms [4][5][6]. For massless states several explicit calculations of scattering amplitude in pure spinor formalism have allowed a direct comparison with the corresponding amplitude calculation in RNS formalism [7][8][9][10]. This direct comparison fixed the relative normalization between the vertex operators in PS and RNS formalism for the massless states. In this paper we extend such explicit direct comparison for the first massive states (m 2 = 1 α ) of open superstring. We explicitly compute all possible tree level 3 point JHEP12(2018)071 functions for a fixed ordering of vertex operators in PS formalism involving 2 massless and 1 massive state and compare them directly with their RNS counterparts. The massless spectrum contains a bosonic gluon field a m (with 8 degrees of freedom) and a fermionic gluino field χ α (with 8 degrees of freedom). The first massive states include a bosonic 3 form field b mnp (with 84 degrees of freedom), a bosonic symmetric rank 2 tensor field g mn (with 44 degrees of freedom) and a fermionic spin 3 2 field ψ mα (with 128 = 84+44 degrees of freedom). The computation of the above mentioned 3 point functions in PS formalism requires the massless unintegrated vertex and its covariant θ expansion [1,[11][12][13] along with the massive unintegrated vertex and its covariant θ expansion [14,15]. While the tree level amplitude prescription in PS is straightforward, one typically encounters a large number of terms in θ expansion of massive vertex and the calculation becomes cumbersome to compute solely by hand. For this reason, the computation was done using Cadabra [16,17] and benchmarked by computing 3-point massless-massless-massless amplitudes which are already known in literature (see, e.g. [18]). For use of Cadabra in pure spinor calculations, also see [19]. The final amplitudes were computed directly using this code and compared with their corresponding RNS results. After fixing the normalization of the three massive vertex operators by comparing three correlators (for three massive states), we find that the rest of the correlators agree perfectly. Moreover, while doing the calculations, we have not fixed the positions of the vertex operators to some specific values but evaluated the correlation functions for general positions x 1 , x 2 , x 3 and have checked that their dependence drops out in the final result which is expected from the SL(2, R) invariance. The rest of the paper is organized as follows. In section 2, we review the tree level amplitude prescription in PS. In section 3, we summarize the θ expansion procedure for massive states given in [15] and give the result for the θ expansion to the relevant order we need in this work. Section 4 gives the detailed derivation and results of 3-point amplitudes in PS formalism. In section 5, we compare our results obtained in PS formalism directly with the amplitudes computed using RNS formalism. We conclude in section 6 with some brief discussion. In appendix A, we give our conventions for the PS formalism. In appendix B, we give our conventions for the RNS calculations and summarize the results for 3-point functions in the RNS formalism. Finally, in appendix C, we give the theta expansion of the massless vertex operator in our conventions. 2 Tree amplitude prescription in pure spinor formalism Both minimal as well as the non minimal pure spinor formalisms give the same amplitude prescription at tree level. In this section, we shall review this prescription [1,2]. In particular, we shall focus on 3-point functions on the disk with a specific ordering of vertex operators on the boundary. To evaluate any 3-point correlator, the knowledge of the unintegrated form of the vertex operator for external states is sufficient. All the amplitudes of interest in this paper are of the form where V 1 V 2 V 3 denotes 3-point function on the disk with a fixed ordering. JHEP12(2018)071 In this paper, V 1 and V 2 will be taken to be unintegrated vertex operators creating massless states (gluon or gluino) and V 3 will be the unintegrated vertex operator creating a massive state (b mnp , g mn or ψ mα ). However, we must emphasize, none of the strategy that we shall outline below is particular to this specific kind of 3-point functions. The tree-level amplitude prescription will be equally valid for any 3-point functions of open strings states (massive or massless). Our choice of normalization of pure spinor measure is the standard one in the literature The process of 3-point amplitude computation can be succinctly summarized in a series of steps that is given below. • Step 1: assume a particular order of the vertex operator inserted on the disk and make use of the OPEs between various conformal weight 1 objects to reduce the three point function to a correlation function involving only the three superfields and three pure spinor ghost λ α coming from the three vertices. The correlation function which contains ∂θ α terms do not contribute since they always fail to have the correct number of θ zero modes to give non-zero answer as required by 2.2. Now perform the θ expansion for each of the vertex, viz. V 1 , V 2 and V 3 , explicitly. Although one can do the full θ expansion, but that is rendered redundant due to 2.2 which states that only terms involving exactly five θs can give a non-zero contribution. The relevant order of θ expansion for each of the vertex for a given amplitude can therefore be deduced from this consideration. • Step 2: from the product V 1 V 2 V 3 , retain only the terms which have precisely five θs as all other terms will give zero contribution trivially due to 2.2. 1 • Step 3: expand the physical fields appearing in various vertices in a basis of plane waves with polarizations as the coefficients Here h β 1 ...β k p 1 ...pn are the constant tensor-spinor of appropriate index structure denoting the polarizations for the physical field H β 1 ...β k p 1 ...pn (X). The correlation function V 1 V 2 V 3 at this stage factorizes completely into two separate correlation functions, one involving X m fields and the other involving pure spinor fields. These can be evaluated independently and their result can be multiplied to obtain the final answer for a given ordering. • Step 4: evaluate the 3-point function on disk by the usual methodology of bosonic open strings. Typically these correlation functions can be put into the schematic form : e ik 1 ·X(x 1 ) :: e ik 2 ·X(x 2 ) :: 1 Notice that 2.2 also suggests we should retain only terms with exactly three number of λs. But this is automatically ensured by the fact that by construction all unintegrated vertex in PS formalism have ghost number 1 and therefore they each carry exactly a single factor of λ. Consequentially the product V1V2V3 always goes as λ 3 . JHEP12(2018)071 • Step 5: the correlators living on pure spinor superspace can be evaluated following the list of identities first derived in [8,20]. We list in appendix A.3 the subset of those identities that were needed in evaluating the 3-point functions in our case. • Step 6: multiply the results obtained from step 4 and step 5 to obtain the full contribution for a given order in theta expansion of the 3 vertices. Once the answer has been obtained for a given order, the final answer for all other inequivalent orders can be readily obtained by suitable permutation of momenta and polarization labels. Since each ordering carries different Chan-Paton factors, the full amplitude cannot be obtained by simply adding the contribution coming from different ordering before multiplying them by Chan-Paton factors. Therefore, to compare our result with the result obtained in RNS formalism, we shall directly compare each inequivalent ordering separately. While the algorithm described above is quite straightforward in principle, the growing number of terms in θ expansion of the vertex operators implies that the cleanest way to perform these amplitude calculations beyond a point is to employ the help of an available computer algebra system. As mentioned in the introduction, we used Cadabra [16,17] which is an open source computer algebra system developed to aid in field-theoretic computations. For our present purpose, we found Cadabra to be unparalleled in implementing the algorithm described above. The bosonic degrees of freedom are contained in a traceless symmetric tensor g mn and a 3-form field b mnp satisfying First massive vertex operator and its θ expansion The constraints (3.1) and (3.2) ensure that the number of independent components in the fields ψ mβ , b mnp and g mn are 128, 84 and 44 respectively. These fields form a massive spin-2 supermultiplet in 10 dimensions. In pure spinor formalism, we use the language of superspace to describe the system in a manifestly supersymmetric invariant manner. This is done by introducing three basic superfields Ψ mα , B mnp and G mn whose theta independent components are ψ mα , b mnp and g mn respectively. These basic superfields satisfy the superspace equations 2 [15] D α G sm = 16 ∂ p (γ p(s Ψ m) ) α (3.3) 2 We use the same conventions as in the reference [15]. Thus, e.g., our convention for the antisymmetrization and symmetrizations are (3.5) and the constraints (3.4) and (3.5) give recursion relations to determine the complete theta expansion of the superfields B mnp , G mn and Ψ mα [15]. Before giving the theta expansion results, we note the unintegrated vertex operator for the 1st massive states of the open string which can be expressed in terms of the basic superfields as [14] JHEP12(2018)071 where [14,15], For calculating any amplitude involving the massive states in pure spinor formalism, we shall need the theta expansion result of the above vertex operator. Clearly, the theta expansion of the basic superfields B mnp and Ψ mα automatically implies the theta expansion of the full vertex operator. For our purposes, we shall need the theta expansion results of Ψ mα upto order θ 3 and that of B αβ upto order θ 4 . We used the mathematica package GAMMA to do this computation [21]. Using equations (3.3), (3.4) and (3.5), we obtain (for details, please see [15]) JHEP12(2018)071 Similarly, the θ expansion of the superfield B αβ is given by where, b mnp = e mnp e ik·X , e mn = e mn e ik·X , ψ mα = e mα e ik·X (3.11) Using these theta expansion results in (3.7) we get the theta expansion of the unintegrated vertex operator. Three point functions using pure spinor formalism In this section, we calculate the 3-point functions involving two massless and one massive state using the pure spinor formalism. In subsection 4.1, we simplify the pure spinor correlators and set up the problem at the superfield level and in subsection 4.2, we evaluate the resulting correlators. Simplifying 3-point correlator Since we are only interested in the 3-point functions on the disk, the location of all the vertex operators can be fixed and hence we only need to consider the unintegrated vertex JHEP12(2018)071 operators. Thus, the correlator we need to evaluate is given by where, V b,g,ψ denotes the massive vertex operator (3.7) and V A denotes the unintegrated vertex operator of the massless states given by 3 The unintegrated vertex operators in (4.1) are fixed at some arbitrary locations x i . The SL(2, R) invariance on the disc guarantees that the 3-point function is independent of the choices of x i . Using the expressions of the unintegrated vertex operators, the desired 3-point function is given by We have suppressed the superscript 3 from the massive superfields since there is only one massive state and hence there is no chance of any confusion. We now manipulate each term of (4.3) one by one. First term. The first term is given by where, the superfields with tilde denote the same superfields with e ik·X factor stripped off. Thus, e.g., in the theta expansion ofB mnp , we just write the polarization tensor e mα , e mnp and e mn instead of ψ mα , b mnp and g mn respectively. We have also used the momentum conservation to write The T 1 does not contribute to any of the amplitude since it does not provide the 5 zero modes of θ α which is required for the non vanishing of pure spinor correlators. Second term. The 2nd term is given by To simplify this term, we use the OPE of d α with superfields to obtain In going to the last line, we have unwrapped the contour to enclose the points x 1 and x 2 . This gives a sign. A further sign comes while moving d α across A α . The signs in front of the individual terms in the last line are net effect of these sign factors. We now use equations (3.4), (3.8) and (C.1), the identity QV = α 2 λ α D α V for an arbitrary superfield V and unwrap the contour of Q. After using the on-shell condition and the pure spinor fierz identity we obtain (after some simplification using the gamma matrix identities) and Third term. The 3rd term is given by Again, in going to the 3rd equality, we have unwrapped the contour and used the OPE between Π m and superfields. In going to the last line, we have performed the contour integration and used the momentum conservation and the identity (4.7). JHEP12(2018)071 Fourth term. Finally, the 4th term is given by 3iα 64 In going to the 3rd equality, we have unwrapped the contour and used the OPE of N mn with λ α whereas in going to the 4th equality, we have used the expression of F αmn in terms of Ψ mα as given in (3.8). We note that T 4 is exactly minus of theT 2 as given in (4.9). Combining all the terms, we find that the total 3-point function is given by where T 2 and T 3 are given in equations (4.10) and (4.11) respectively. Below, we shall give the results for the correlators V 1 V 2 V 3 for different choices of the 2 massless and one massive external states. Evaluation of correlators The two terms T 2 and T 3 given in (4.13) are at the superfield level. To compute some specific 3-point function, we need to keep only the fields of interest to be non zero in the superfields. After specializing to some specific amplitude, A 3 can be evaluated using the theta expansion results given in section 3 and appendix C and the pure spinor correlators listed in appendix A.3. We use the symbolic computer programme Cadabra to do this calculation [16,17]. An important point to note is that T 2 as given in equation (4.10) depends upon the world-sheet coordinates x i but T 3 does not. However, the dependence of T 2 on the world sheet coordinates automatically drops out on evaluating it using the theta expansion of the superfields as expected from the SL(2, R) invariance. We now give the results for different 3-point correlators. 2 gluon and 1 b mnp field. For this case, we have This gives e mnp e 1 m e 2 n (k 1 ) p (4.14) JHEP12(2018)071 2 gluon and 1 g mn field. For the aag amplitude, we have This gives, 2α (e 1 · k 2 )(e 2 · g · k 1 ) + 2α (e 2 · k 1 )(e 1 · g · k 2 ) −2α (e 1 · e 2 )(k 1 · g · k 2 ) + (e 1 · g · e 2 ) (4.15) 2 gluino and 1 b mnp field. For the χχb amplitude, we have This gives 2 gluino and 1 g mn field. For the χχg amplitude, we have This gives, 1 gluon, 1 gluino and 1 ψ mα field. In this case, since all the external states are different, the two orderings aχψ and χaψ are different. We give the result for both cases. For the aχψ correlator, we have This gives On the other hand, for the χaψ correlator, we have This gives JHEP12(2018)071 3 massless fields. For comparing the pure spinor results with the corresponding RNS results, we also need the massless amplitudes. For aaa correlator, the pure spinor calculation gives the following result in our conventions aaa = i 180 (e 1 · e 2 )(e 3 · k 1 ) + (e 1 · e 3 )(e 2 · k 3 ) + (e 2 · e 3 )(e 1 · k 2 ) (4.20) On the other hand, for the aχχ correlator, we get From the explicit expression of all massless-massless-massive amplitudes obtained in this section, we observe that all such 3-point functions are symmetric under a change of cyclic order. Therefore the inclusion of Chan-Paton factors (denoted by t a , t b t c ) will make the full amplitude proportional to T r(t a , {t b , t c }). Compare this with the 3-point functions involving all massless states, which are antisymmetric under a change of cyclic order and therefore after taking into account the Chan-Paton factors, the final answer becomes proportional to T r(t a , [t b , t c ]). Another point to note is that all the amplitudes considered in this section are invariant under the gauge transformations e i → e i + k i . We now turn to comparing the RNS and PS results. Comparing pure spinor and the RNS results By comparing the pure spinor results given above with the corresponding RNS results given in appendix (B.2), we see that the tensor structures of the 3-point functions match perfectly. Moreover the relative coefficients of the various terms in the correlators aag and aχψ which have more than one terms, also match exactly. This is a non trivial test. We shall now show that the overall numerical factors (i.e. normalizations) of the different 3-point functions in pure spinor and RNS are also in perfect agreement with each other. We have denoted the overall normalization of the vertex operators in the RNS calculations relative to those in PS calculations by g a , g χ etc. (see appendix B.2). For example, if N RNS and N PS denote the normalizations of the gluon vertex operator in the RNS and PS respectively, then for the V a V a V a correlator, we have (denoting V ≡ NṼ ) From this, it is clear that only the relative normalization between RNS and PS vertex operators have any physical significance. To exploit this fact, we define In our calculations, we have set the overall normalization of the PS vertex operators to be 1 and kept the relative normalization factor g a , g χ etc. to be in the RNS vertex operators. With the above convention, the overall RNS and the pure spinor numerical factors for each correlator is given in the table 1. By comparing the RNS and pure spinor numerical factors for aaa and aχχ , we find In terms of g a and g χ , the numerical factors for aab , aag and aχψ give The above values of g b , g g and g ψ agree perfectly with the value obtained using χχb , χχg and χaψ correlators. This is a non trivial consistency check. Discussions In this paper we have explicitly shown the equivalence of all massless-massless-massive 3-point functions for each given ordering of vertex operators on the boundary for open superstrings computed in PS and RNS formalisms. For the massive states, the normalization factors, which are fixed, using say aab , aag and aχψ correlators, match perfectly with the χχb , χχg and χaψ correlators which establishes the consistency of the relative normalization determined in this paper and establishes firmly the equivalence between RNS and PS formalism for first massive states. Once the vertex for open superstring is known, its extension to closed strings can be obtained in a straightforward manner by taking appropriate tensor products of left and right moving sectors [22]. To compare loop amplitudes one needs to compute the amplitude in PS formalism using integrated vertex constructed for massive states in [23]. With the relative normalization now fixed, they should match exactly with the corresponding RNS result. Also as argued in [23], the strategy behind constructing massive vertex (integrated or unintegrated) is not sensitive to mass level in question and is expected to be identical for all higher massive states. Therefore, one can reasonably expect that all 3-point functions involving higher massive states can also be directly compared with RNS results and yield a consistent set of normalizations for each mass level. JHEP12(2018)071 Acknowledgments We are deeply thankful to Ashoke Sen for many insightful discussions throughout the course of this work and for useful comments on the draft. We would also like to thank Anirban Basu, Rajesh Gopakumar and Satchitananda Naik for their encouragement and comments on the draft. SPK is also thankful to the organizers of Strings 2018 held in Okinawa Institute of Science and Technology (OIST), Japan for giving him the opportunity to give a gong show on a preliminary version of this work. MV is also thankful to Institute of Theoretical Physics, Chinese Academy of Sciences, China and the OIST, Japan for hospitality while this work was in progress. This research was supported in part by the Infosys Fellowship for the senior students. We also thank the people and Government of India for their continuous support for theoretical physics. A Some pure spinor results In this appendix, we note some pure spinor results which are used in this work. A.1 Pure spinor conventions In this section, we give our conventions for the pure spinor calculations. For the tree amplitude calculations, both the minimal as well as the non minimal formalisms give the same prescription. For convenience, we shall only consider the minimal formalism. The world sheet action of the minimal formalism is given by where, m = 0, 1, , · · · , 9 and α = 1, · · · , 16. The p α and w α are the conformal weight one conjugate momenta of the conformal weight zero fields θ α and λ α respectively. The fields with upper (lower) spinor indices transform as right (left) handed Weyl spinor. The λ α satisfy the pure spinor constraint The BRST operator is defined to be Due to pure spinor constraint, the world-sheet fields can only appear in the following gauge invariant combinations 4 The coincident operators are normal ordered via where, A and B are any two operators and the contour surrounds the point z. However, we shall suppress the normal ordering symbol : : throughout this draft. JHEP12(2018)071 Any other gauge invariant combination can be expressed in terms of these objects. The ghost current is given by J = λ α w α which implies that the λ α carries the ghost number +1 and the rest of the fields carry zero ghost number. Before proceeding ahead, we note down some important OPEs of the theory which arise frequently in calculations In the above OPEs, ∂ m is the derivative with respect to the spacetime coordinate X m , ∂ is the derivative with respect to the world-sheet coordinate, V denotes an arbitrary superfield. The d α and Π m denote the supersymmetric combinations The D α is the supercovariant derivative given by For calculating the scattering amplitudes, we need to know the number of zero modes of the world-sheet fields. Due to the pure spinor constraints, it follows that all the worldsheet ghost sector fields, namely λ α and w α have 11 independent components. Now, it is a result from the theory of Riemann surfaces that the number of zero modes of the conformal weight one and conformal weight zero objects on a genus g Riemann surface is g and 1 respectively. Thus, the number of zero modes of λ α on a genus g Riemann surface is 11 whereas the number of zero modes of w α on a genus g Riemann surface is 11g. On the other hand, the matter sector fields θ α and p α have 16 and 16g zero modes respectively on a genus g Riemann surface. As in RNS formalism, the integration over the zero modes give divergences and we need some way to absorb these zero modes. For details, see the original literatures. Before moving further, we describe an important identity on the world sheet which is needed while solving the BRST equations of motion. The open string vertex operators live on the boundary, i.e., along the real axis on the complex plane. This implies that in the BRST equation QU = ∂ R V , the derivative in the right hand side is along the real axis (and hence the subscript R in the right hand side). However, for doing calculations, it is JHEP12(2018)071 convenient to convert it into the holomorphic derivative using the chain rule. If x denotes the coordinate along the real axis, then for an arbitrary function f on the real axis, we have Now, using the chain rule, we have Using the equation of motion for p α , namely,∂θ α = 0 and the above two equations, we obtain Finally, using (A.6) and (A.7), we can express this as In the above derivation, we have used the world sheet fields defined using the doubling trick. For a derivation without using doubling trick and more details, please see the section 2 of [23]. A.2 Hodge duality of gamma matrices in d=10 In 10 dimensions, the 16 × 16 gamma matrices satisfy the following Hodge dualities where, m 1 ···m 9 is the 10 dimensional epsilon tensor defined as Due to the above dualities, not all the antisymmetrized product of gamma matrices are independent. We shall take γ m 1 , γ m 1 m 2 , γ m 1 m 2 m 3 , γ m 1 m 2 m 3 m 4 and γ m 1 m 2 m 3 m 4 m 5 along with the identity matrix I 16×16 as the linearly independent basis elements for vector spaces of 16 × 16 complex matrices. A.3 Pure spinor superspace identities While computing the scattering amplitudes in pure spinor formalism, the last step requires evaluation of integration over the zero modes of λ α and θ α . Due to the pure spinor constraints and the symmetry properties of θ α and λ α , there are only a finite number of these basic pure spinor correlators which are non zero. Below, we list those pure spinor correlators which are used in this note [8] (λγ m θ)(λγ n θ)(λγ p θ)(θγ stu θ) = 1 120 (A.21) B RNS results In this section, we summarize the results of the 3-point functions involving two massless and one massive states computed using the RNS formalism. In subsection B.1, we state our conventions and in subsection B.2, we give the results of 3-point computations. A useful reference for this section is [24] (also see [25,26]). B.1 Conventions for RNS calculations Due to the picture number anomaly on a genus g Riemann surface, a non vanishing RNS correlator must have the picture number 2g − 2. This is ensured by working with vertex operators in appropriate picture number and the insertions of appropriate number of picture changing operators (PCOs). For the 3-point functions on Riemann sphere, we can avoid the insertions of PCOs by working with vertex operators of the appropriate picture number so that the total picture number adds up to −2 (which is the picture number anomaly on Riemann sphere). We start by writing down the vertex operators for the massless states in various picture numbers. The gluon vertex operator in the −1 and 0 picture numbers is given by e m i∂X m (x) + 2α k n ψ n (x)ψ m (x) e ik·X(x) (B.1) JHEP12(2018)071 The gluino vertex operator in the −1/2 picture number is given by The polarization vectors of the gluon and gluino satisfy the transversality conditions Now, we turn to the vertex operators for the first massive states [27,28]. The vertex operators for the anti-symmetric 3-form field b mnp in the picture numbers −1 and 0 are given by The vertex operators for the symmetric traceless massive graviton field g mn are given by 2α i∂X m i∂X n + ∂ψ m ψ n + i∂X m k p ψ p ψ n e ik·X (B.5) Finally, the vertex operators for the massive gravitino field ψ mα in the −1/2 picture number is given by e ik j ·X(y j ) (B.10) JHEP12(2018)071 This can be evaluated recursively using (B.9). We shall encounter the situation where some w may coincide with some y j . E.g., we shall need the following correlator In going to the 2nd line, we have used the definition of the normal ordering (A.4) and in going to the 3rd line, we have used (B.10) for q = 1 and n = 3. Similarly, we also need Next, we consider the world-sheet correlators involving the ψ m fields This expression can also be evaluated recursively. In the final step, the two point function can be evaluated by using the OPE ψ m (z)ψ n (w) = η mn z − w + · · · (B.13) Next, we consider the ghost sector. The basic correlator involving the reparametrization ghost c(x) is given by (B.14) The basic correlator involving the bosonized ghost field φ(z) is given by Finally, we consider the spin fields. The basic OPEs involving the spin fields are given by (a consistent set of these OPEs can be found in [29]) , j mn ≡ : ψ m ψ n : (B.21) Using the above OPEs, we can work out the following useful correlators which are needed in our calculations B.2 3-point functions We are now ready to give the results of 3-point functions of two massless and one massive field (for the computation of 3-point functions of states in leading Regge trajectory in RNS, see [30]). These are straightforward to evaluate using the vertex operators and the correlators given in the previous subsection. Hence, we just state the final results below. For our calculations, we shall take the bosonic massless and massive fields to be either in the −1 or 0 picture numbers. On the other hand, the fermionic massless or massive fields will always be taken in the −1/2 picture. The picture numbers will be shown by a superscript on the vertex operators inside correlators. Thus, the 3-point amplitudes are given by We start by considering the 3-point functions involving all massless fields. The two possible correlators in this case are aaa and aχχ . Using the world-sheet correlators given above, these 3-point functions can be evaluated to be 2α (e 1 · e 2 )(e 3 · k 1 ) + (e 1 · e 3 )(e 2 · k 3 ) + (e 2 · e 3 )(e 1 · k 2 ) (B.25) As an amplitude this vanishes on summing (k 1 ↔ k 2 ) term if the gauge group is abelian. Similarly, the aχχ correlator can be worked out to be Again, for the abelian gauge group, the corresponding amplitude vanishes. JHEP12(2018)071 Next, we consider the two massless and one massive field. We start with two gluons and one b mnp field Next, we consider the 3-point function of two gluons and one massive g mn field which is given by = −g 2 a g g 2α η mn e 1 m e 2 n e pq k p 1 k q 1 + e 1 p e 2 q e pq + 2α e 1 m e 2 q e pq k p 1 k m 2 − 2α e 1 q e 2 m e pq k m 1 k p 1 (B.28) Next, we include gluino and consider the χχb and χχg correlators which can be worked out to be and, Finally, we consider the massive fermion. The two different 3-point functions with one massive fermion and two massless fields are aχψ and χaψ which are given by and, To compare the results involving the fermionic fields with the PS results, we need to first convert the RNS gamma matrix conventions into the PS gamma matrix conventions. This mainly involves setting the charge conjugation matrix to be the Kronecker delta δ α β which implies (for details, see e.g., [24,25]) and so on. JHEP12(2018)071 C Theta expansion of massless vertex operator Theta expansion of the massless vertex operator is known and has been extensively used in the literature (see, e.g., [18]). However, some of our conventions (e.g., equation (A.7)) are different from the literature in which the theta expansion of massless vertex operator is used. Below, we derive the results in our convention briefly indicating the steps. We start by recalling the N = 1 SYM equations in 10 dimensions [31]. N = 1 SYM equations in 10 dimensions. In 10 dimensions, the open string massless states are described by the 10 dimensional N = 1 super Yang Mills equations. The field strengths describing the theory are given by . The 10 dimensional Yang-Mills equations of motion follow from Using this along with the Bianchi identities, we obtain the following equations at linearized level Further, the superfields A m , W α and F mn can be expressed as We also have, where we have used the pure spinor open string identity (A.11) in going to the second line. The BRST equation of motion now gives, To match these equations with the 10 dimensional SYM equations, we rescale the fields as After this rescaling, the BRST equation gives The first 4 equations are precisely satisfied by the 10 dimensional N = 1 SYM equations given in appendix C whereas the last equation is satisfied by the pure spinor constraint. The correctly normalized vertex operators of the massless states in our conventions thus become Theta expansion of massless superfields. Finally, we turn to the theta expansion. We shall follow the steps outlined in [13]. We shall need the following equations for doing the theta expansion JHEP12(2018)071 Before proceeding to do the theta expansion, we need to fix a gauge. We shall choose the gauge θ α A α = 0. In this gauge choice, we have Now, multiplying by θ β in the 1st equation and by θ α in the 2nd and 3rd equations of (C.9) and using the above identity along with the gauge choice θ α A α = 0, we obtain (1 + D)A α = 2A m (γ m θ) α , DA m = −(θγ m W ) , DW α = 1 2 F mn (γ mn θ) α (C.11) where we have defined D ≡ θ α D α = θ α ∂ α . We can use the above 3 equations along with the 4th equation of (C.9) to do the theta expansion. If we denote the th order component of the superfield M by M ( ) , then we have M ( ) = m α 1 ···α θ α 1 · · · θ α =⇒ DM ( ) = M ( ) (C.12) Using this, the 3 equations of (C.11) and the 4th equation of (C.9) give the following recursive relations Denoting the theta independent components of the superfields A m and W α to be A (0) m ≡ a m , W α (0) ≡ χ α (C.14) the above recursive relations give where f mn = ∂ m a n − ∂ n a m and the plane wave expansion of the gluon and gluino are given by a m = e m e ik·X , χ α = ξ α e ik·X (C. 16) Open Access. This article is distributed under the terms of the Creative Commons Attribution License (CC-BY 4.0), which permits any use, distribution and reproduction in any medium, provided the original author(s) and source are credited.	2023-01-21T14:29:42.178Z	2018-12-01T00:00:00.000	{ "year": 2018, "sha1": "fdefefa0d3418c9ac4626d065718f3f203841b25", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/JHEP12(2018)071.pdf", "oa_status": "GOLD", "pdf_src": "SpringerNature", "pdf_hash": "fdefefa0d3418c9ac4626d065718f3f203841b25", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [] }
1424642	pes2o/s2orc	v3-fos-license	Long-Term Once-Daily Tiotropium Respimat® Is Well Tolerated and Maintains Efficacy over 52 Weeks in Patients with Symptomatic Asthma in Japan: A Randomised, Placebo-Controlled Study Background This study assessed the long-term safety and efficacy of tiotropium Respimat, a long-acting inhaled anticholinergic bronchodilator, in asthma, added on to inhaled corticosteroids (ICS) with or without long-acting β2-agonist (LABA). Methods 285 patients with symptomatic asthma, despite treatment with ICS±LABA, were randomised 2:2:1 to once-daily tiotropium 5 μg, tiotropium 2.5 μg or placebo for 52 weeks (via the Respimat SoftMist inhaler) added on to ICS±LABA, in a double-blind, placebo-controlled, parallel-group study (NCT01340209). Primary objective: to describe the long-term safety profile of tiotropium. Secondary end points included: trough forced expiratory volume in 1 second (FEV1) response; peak expiratory flow rate (PEFR) response; seven-question Asthma Control Questionnaire (ACQ-7) score. Results At Week 52, adverse-event (AE) rates with tiotropium 5 μg, 2.5 μg and placebo were 88.6%, 86.8% and 89.5%, respectively. Commonly reported AEs with tiotropium 5 μg, 2.5 μg and placebo were nasopharyngitis (48.2%, 44.7%, 42.1%), asthma (28.9%, 29.8%, 38.6%), decreased PEFR (15.8%, 7.9%, 21.1%), bronchitis (9.6%, 13.2%, 7.0%), pharyngitis (7.9%, 13.2%, 3.5%) and gastroenteritis (10.5%, 3.5%, 5.3%). In the tiotropium 5 μg, 2.5 μg and placebo groups, 8.8%, 5.3% and 5.3% of patients reported drug-related AEs; 3.5%, 3.5% and 15.8% reported serious AEs. Asthma worsening was the only serious AE reported in more than one patient. At Week 52, adjusted mean trough FEV1 and trough PEFR responses were significantly higher with tiotropium 5 μg (but not 2.5 μg) versus placebo. ACQ-7 responder rates were higher with tiotropium 5 μg and 2.5 μg versus placebo at Week 24. Conclusions The long-term tiotropium Respimat safety profile was comparable with that of placebo Respimat, and associated with mild to moderate, non-serious AEs in patients with symptomatic asthma despite ICS±LABA therapy. Compared with placebo, tiotropium 5 μg, but not 2.5 μg, significantly improved lung function and symptoms, supporting the long-term efficacy of the 5 μg dose. Trial Registration ClinicalTrials.gov NCT01340209 Introduction Asthma is a chronic inflammatory disease affecting approximately 300 million people worldwide, and has a prevalence rate of around 7% in Japan [1,2]. Asthma treatment aims to reduce symptoms and exacerbation risk by treating the characteristic airway inflammation and bronchoconstriction, enabling patients to lead a normal and healthy life [1,3]. Inhaled corticosteroids (ICS) provide the cornerstone of asthma prophylactic treatment, supplemented by rapid-onset, short-acting bronchodilators for when fast symptomatic relief is required. Global Initiative for Asthma guidelines recommend a step-wise approach to treatment that relies on a continuous cycle of assessment, treatment and monitoring [1]. Low-dose ICS are recommended as the initial controller treatment. If symptoms are not adequately controlled, Global Initiative for Asthma guidelines recommend adding a long-acting β 2 -agonist (LABA), such as formoterol or salmeterol, before physicians consider increasing the dose of ICS or adding sustained-release theophylline or a leukotriene modifier. However, despite currently available therapies and detailed guidelines, at least 40% of patients with asthma have symptomatic or poorly controlled disease [4][5][6]. There remains a need to expand the therapeutic options available for patients whose asthma is not adequately controlled with ICS with or without a LABA. Tiotropium is a once-daily long-acting anticholinergic bronchodilator currently approved for maintenance treatment in patients with chronic obstructive pulmonary disease (COPD). In Phase II and III placebo-controlled trials, tiotropium Respimat (Boehringer Ingelheim, Ingelheim am Rhein, Germany) add-on to high-dose ICS and LABA improved lung function versus placebo in patients who were still symptomatic despite treatment with ICS plus a LABA [7]. Tiotropium Respimat add-on also reduced the risk of asthma exacerbation and asthma worsening versus placebo [8]. Furthermore, the addition of tiotropium HandiHaler (Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, Connecticut, USA) has been shown to be non-inferior to the addition of salmeterol [9,10] and superior to doubling the dose of ICS [10]. Tiotropium has demonstrated a favourable safety profile, with rates of adverse events (AEs) in patients with symptomatic asthma comparable with placebo when added to maintenance therapy in randomised clinical trials [7][8][9][10]. We conducted a Phase III, randomised, double-blind, placebo-controlled, parallel-group study to evaluate the long-term safety and efficacy of tiotropium Respimat (5 μg or 2.5 μg) compared with placebo Respimat in patients from Japan whose asthma was not being adequately controlled by ICS with or without a LABA. Study design The protocol for this trial and supporting CONSORT checklist are available as supporting information; see S1 CONSORT Checklist and S1 Protocol). In this Phase III, randomised, doubleblind, placebo-controlled, parallel-group study (ClinicalTrials.gov identifier: NCT01340209; http://www.clinicaltrials.gov/ct2/show/NCT01340209?term=NCT01340209&rank=1), patients entered a 4-week screening period before being randomised to receive once-daily tiotropium 5 μg (two actuations of 2.5 μg), tiotropium 2.5 μg (two actuations of 1.25 μg) or placebo for 52 weeks, all delivered via the Respimat SoftMist inhaler (Boehringer Ingelheim) as add-on to ICS with or without a LABA. Patients were followed up for an additional 3 weeks following completion of the randomised treatment period. Patients were required to visit the trial centre every 4 weeks during the treatment period to record AEs and drug accountability. Lung function was assessed at initial screening, randomisation, Week 12, Week 24, Week 36 and Week 52 (at the end of treatment); these visits were conducted in the evening whereas all other visits could be conducted during the day. During the screening and treatment periods, patients used an electronic diary to record their asthma symptoms, quality of life, use of rescue medication and administration of the study medication. Patients who withdrew prematurely from the randomised treatment period continued to be followed up for their vital status. The trial was carried out in compliance with the Declaration of Helsinki and in accordance with the International Conference on Harmonisation Harmonised Tripartite Guideline for Good Clinical Practice. Before initiation, the trial protocol, patient information sheet and consent form were reviewed and approved, according to national and international regulations, by the Japanese competent authority and the institutional review board of each participating centre: Obihiro Kokyukika Naika Hospital; Hitachi, Ltd. Participants Outpatients with symptomatic asthma lasting 12 weeks or more, and who were aged 18-75 years, were eligible for enrolment. The initial diagnosis of asthma must have been made before the age of 40 years, and was confirmed at screening with a bronchodilator reversibility (15-30 minutes after 400 μg salbutamol) resulting in a forced expiratory volume in 1 second (FEV 1 ) increase of 12% and 200 mL. Patients were required to have been receiving maintenance treatment with stable medium-dose ICS (eg 400 μg and 800 μg budesonide or equipotent dose), alone or in a fixed combination with a LABA, for at least 4 weeks prior to screening. Symptomatic asthma was defined as a seven-question Asthma Control Questionnaire (ACQ-7) score of 1.5 at screening and prior to randomisation. All patients must also have had prebronchodilator FEV 1 60% and 90% of predicted normal, and to have never smoked or be ex-smokers who stopped smoking at least 1 year prior to enrolment, with a smoking history of less than 10 pack-years. The main exclusion criteria were COPD and other significant unstable concomitant disease. Patients were not permitted to take any investigational drug, non-topical beta-blockers or other asthma therapies within 4 weeks prior to enrolment and/or during the screening period, or anti-immunoglobulin E antibodies within 6 months of enrolment and/or during the screening period. Patients were also excluded if they: had any asthma exacerbation or any respiratory tract infection within 4 weeks prior to enrolment and/or during the screening period; were participating in another trial; had narrow-angle glaucoma and/or micturition disorder because of prostatic hyperplasia; and had failed to complete 80% of their electronic diary during the run-in period. Randomisation and treatment Eligible patients were randomised in blocks, 2:2:1, to once-daily add-on tiotropium Respimat 5 μg, tiotropium Respimat 2.5 μg or placebo Respimat for 52 weeks. The 2:2:1 randomisation ratio was employed due to ethical considerations, as it would reduce the number of patients receiving placebo Respimat. Each patient received a single randomisation number indicating their allocated treatment. Randomisation was achieved via a third-party phone-or web-based system involving a validated pseudo-random number generator and a supplied seed number, using a block size of 5. All patients, including those randomised to the placebo Respimat group, continued to receive background medium-dose ICS with or without a LABA. Study medication was taken each evening when patients inhaled two actuations using the Respimat SoftMist inhaler. In order to maintain the blind, patients in the placebo group also used the Respimat SoftMist inhaler, and the placebo inhalation solution was identical in appearance to the tiotropium inhalation solution. Blinding was maintained until after database lock. Administration of rescue medication was allowed at any point during the trial. Open-label salbutamol hydrofluoroalkane metered-dose inhaler (100 μg per actuation) was provided as rescue medication for use as needed. Formoterol, alone or in fixed combinations with ICS, was not allowed as rescue medication in this trial. Temporary addition of systemic corticosteroids and theophylline preparations, temporary increases in the dose of ICS and the use of antibiotics were also allowed, at the discretion of the investigator, to control acute asthma exacerbations. Study end points The primary objective of this trial was to evaluate the long-term safety of two doses (5 μg and 2.5 μg) of tiotropium Respimat, compared with placebo Respimat, as add-on to maintenance therapy with medium-dose ICS with or without a LABA, in patients with symptomatic asthma that was not adequately controlled at baseline. Safety was assessed by recorded AEs, vital signs (pulse rate and blood pressure), laboratory data and electrocardiogram. Secondary efficacy end points included: trough FEV 1 and trough forced vital capacity responses; in-clinic trough peak expiratory flow rate (PEFR) response using a spirometer; and ACQ-7 responder rate (minimal important difference of 0.5). Lung function responses were defined as change from baseline to the stated week; baseline was the pre-treatment value recorded at the randomisation visit 10 minutes before the evening dose of the patient's usual ICS controller medication and first dose of study medication. Statistical analyses The sample size was determined with reference to the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use E1 guideline on population exposure, to assess long-term clinical safety and to satisfy Japanese regulatory authority requirements. At least 100 patients were required to remain on treatment for 52 weeks to describe the safety profile of tiotropium Respimat 5 μg and 2.5 μg. Assuming a drop-out rate of 10%, it was calculated that 280 patients (112 patients per tiotropium Respimat group and 56 patients in the placebo Respimat group) should be enrolled. The summary of safety data was based on the treated set (all randomised patients who received one or more doses of study medication) and was evaluated using descriptive analysis. The study was not designed to detect any difference between the two doses of tiotropium Respimat and placebo Respimat in terms of overall or any specific AE incidence. The placebo Respimat group was included to obtain information on the safety profile of the placebo-control group (on a background of ICS with or without a LABA) and to reduce any possible reporting bias for drug-related AEs. The analysis of efficacy data was performed on the full analysis set (all treated patients with baseline data and at least one on-treatment efficacy measurement). Power calculations for these secondary analyses were not conducted, as the primary objective of the study was to assess the safety profile of tiotropium Respimat. Lung function tests were evaluated using a restricted maximum likelihood-based mixed effects model with repeated measures. Analyses included the fixed, categorical effects of 'treatment', 'visit' and 'treatment-by-visit interaction', as well as the continuous, fixed covariates of 'baseline value' and 'baseline value-by-visit interaction'. 'Patient' was included as random effect. Analyses were performed on least squares means using a two-sided α = 0.05 (two-sided 95% confidence intervals [CIs]). As it was not planned to test any statistical hypothesis in a confirmatory manner, all p values were nominally interpreted and no correction for multiple testing was applied. Other end points were summarised using descriptive statistics. Patient disposition and demographics Between April 2011 and April 2013, 515 patients were enrolled from 54 centres in Japan. Following screening, 285 patients were randomised to treatment. As per the study protocol, twice as many patients were randomised to each of the tiotropium Respimat groups compared with the placebo Respimat group. Overall, 21 patients (7.4%) prematurely discontinued trial medication and 264 patients completed the study (Fig 1). Baseline demographic characteristics were generally well balanced across the three treatment groups ( Table 1). The majority of patients were female (61.8%), mean age was 44.5 years and mean body mass index at baseline was 24.4 kg/m 2 . Most patients had never smoked (75.1%); one patient (1.8%) in the placebo Respimat group was a current smoker at screening, which was in violation of the protocol. In terms of lung function at baseline, mean (pre-dose) FEV 1 was 2.28 L and mean percent of predicted FEV 1 was 80.2%. Mean ACQ-7 total score at baseline was 1.95, and at the screening visit mean FEV 1 reversibility was 23%. Overall, mean duration of asthma was 22.4 years (median duration of asthma, 22.0 years) and mean age at onset of asthma was 22.1 years (median age, 24.8 years). Concomitant LABA was used by 57.0% of patients in the tiotropium Respimat 5 μg group, 54.4% of patients in the tiotropium Respimat 2.5 μg group and 61.4% of patients in the placebo Respimat group. Overall, the mean ICS dose of stable maintenance therapy (budesonide equipotent dose) at baseline was 661.7 μg and the median ICS dose was 800 μg (range 400-1600 μg); doses were comparable across the three treatment groups. Long-term safety All randomised patients received one or more doses of trial medication and were therefore included in the treated set for long-term safety analysis. The median exposure (ie duration of treatment) was 365 days in all three treatment groups, and 93.0% of patients had 364 days of exposure. Median treatment compliance was 93.7%. Nine cardiac disorder events were reported in five patients (4.4%) in the tiotropium Respimat 5 μg group, compared with one event in one patient (0.9%) in the tiotropium Respimat 2.5 μg group, and no patients in the placebo Respimat group (Table 4). Five cardiac AEs in the tiotropium Respimat 5 μg group occurred in the same patient at the same time, soon after the start of trial medication, and resolved shortly after discontinuation, which were therefore assessed by the investigator as being drug-related. These events (atrioventricular block [first degree], extrasystoles, palpitations, supraventricular extrasystoles and ventricular extrasystoles) were all mild in intensity and required no treatment. The same patient in the tiotropium Respimat 5 μg group also had hypertension that was considered to be drug-related. Serious AEs (SAEs) were less frequent with tiotropium Respimat 5 μg (n = 4; 3.5%) or tiotropium Respimat 2.5 μg (n = 4; 3.5%) than with placebo Respimat (n = 9; 15.8%). One SAE in each of the tiotropium Respimat 5 μg and placebo Respimat groups was coded as asthma worsening, all other SAEs were single occurrences, and only the asthma worsening event in the placebo Respimat group was considered to be drug-related (S1 Table). No deaths or lifethreatening AEs were reported during the study. Long-term efficacy The full analysis set for the long-term efficacy analysis comprised 114, 114 and 56 patients in the tiotropium Respimat 5 μg, tiotropium Respimat 2.5 μg and placebo Respimat groups, respectively, for whom baseline data and at least one on-treatment efficacy measurement were available. Discussion Data presented here show that once-daily tiotropium Respimat add-on to current ICS maintenance therapy, with or without a LABA, had a safety and tolerability profile that was balanced compared with placebo Respimat in patients from Japan whose asthma was not being adequately controlled at baseline. Rates of SAEs and AEs leading to discontinuation were low. None of the SAEs occurring in patients receiving either dose of tiotropium Respimat was considered to be drug-related. The majority of AEs were mild to moderate in intensity. The study sample size was based on local regulatory requirements, and the study was therefore not formally powered for safety or efficacy. Although this may be a limitation of the study, a placebo Respimat group was included that allows comparison and gives scope to compare the results described here with those obtained in other studies. In addition, the placebo Respimat arm may reduce possible reporting bias for drug-related AEs. Overall, it appeared that rates of AEs were similar between treatment groups, and rates of SAEs were numerically lower in patients receiving tiotropium Respimat compared with those receiving placebo Respimat. Drug-related AEs were slightly more frequent with tiotropium Respimat 5 μg than with tiotropium Respimat 2.5 μg or placebo Respimat; most drug-related AEs were single occurrences and none was considered to be serious. None of the SAEs was considered by the treating physician to be drug-related. The frequency of AEs (>86%) reported in all groups, including the placebo Respimat group, was higher than observed in other Phase II and III trials with tiotropium in populations with moderate to severe asthma [7][8][9]. However, in this study with a small sample size, the overall AE trends for tiotropium Respimat versus placebo Respimat were comparable with other trials, and no new safety signals were observed. The predominant AE in each group was nasopharyngitis, suggesting that the elevated AE rates may have been due to seasonal symptoms such as the common cold [11], or that air pollution may have contributed to respiratory illnesses that were reported as AEs [12]. It should also be noted that studies of long-term asthma treatments in Japanese patients tend to report relatively high AE rates. For example, AE rates were 81% and 90%, respectively, in a long-term safety study of fluticasone alone versus fluticasone combined with vilanterol (ClinicalTrials.gov identifier: NCT01244984), whereas in In-clinic trough PEFR response for patients receiving tiotropium Respimat add-on or placebo Respimat add-on. In-clinic trough PEFR response for patients receiving tiotropium Respimat or placebo Respimat as add-on to maintenance therapy of inhaled corticosteroids, with or without a longacting β 2 -agonist, over the 52-week study period (full analysis set). Tiotropium Respimat and placebo Respimat dosed once-daily in the evening. Baseline mean (standard deviation) at visit 2 (randomisation), L/min: 371.0 (120.2). PEFR, peak expiratory flow rate. a study of fluticasone versus fluticasone/vilanterol combination that was conducted in 11 countries outside of Japan, the overall AE rates were 65% and 63% [13]. Previous research has prompted discussions surrounding the cardiac safety of inhaled anticholinergics in patients with respiratory diseases. For example, a meta-analysis of the cardiovascular risks associated with inhaled anticholinergics in patients with COPD identified an increased risk of cardiovascular death, myocardial infarction or stroke [14]. However, a randomised, placebo-controlled, 4-year trial that evaluated the long-term efficacy and safety of tiotropium given by HandiHaler in nearly 6000 patients with COPD [15] demonstrated reduced cardiac morbidity with tiotropium versus placebo, including a lower risk of serious cardiac events (relative risk: 0.84; p<0.05) such as myocardial infarction (relative risk: 0.71; p<0.05) and congestive heart failure (relative risk: 0.59; p<0.05). More recently, a study in 17,000 patients with COPD has shown that the safety profile of tiotropium Respimat 5 μg is comparable with that of tiotropium HandiHaler 18 μg, with an incidence of serious cardiac disorder AEs of 4.8% versus 4.7%, respectively [16]. In the current study, five patients (4.4%) receiving tiotropium Respimat 5 μg reported nine cardiac events, and one patient receiving tiotropium Respimat 2.5 μg reported a cardiac event. Five cardiac events in the tiotropium Respimat group were considered to be drug-related; these all occurred in the same patient. Although the investigator considered the cardiac events to be related to tiotropium therapy, the patient had co-morbid diagnoses of lumbar pain, hepatitis C virus and anaemia at baseline, and was continuously treated with intravenous peginterferon alfa-2 and ribavirin for hepatitis, both of which have been associated with cardiovascular events [17]. Furthermore, each cardiac event was mild in intensity, and it is worth considering whether the events may have caused morbidity or been detected outside of a clinical trial setting. With regards to efficacy, this trial confirmed the long-term lung function benefits of tiotropium Respimat add-on to ICS with or without a LABA in patients from Japan. At Week 52, statistically greater responses that were in line with results from previous international Phase II and III trials [7][8][9] were observed in trough FEV 1 response and in-clinic trough PEFR response for tiotropium Respimat 5 μg versus placebo Respimat. A previous subgroup analysis of an international tiotropium trial in patients with COPD demonstrated that the pattern of treatment benefit from tiotropium was similar in patients from Asia versus the global population [18]. Therefore, tiotropium Respimat 5 μg may be considered appropriate for long-term therapy. The magnitude of improvement in lung function seen in the study presented here with the addition of tiotropium Respimat appears to be of a similar degree to the addition of a LABA, based on a Cochrane review of inhaled LABAs versus placebo on a background of ICS, which reported improvements in FEV 1 and evening PEFR of 110 mL and 17.89 L/min, respectively, with LABAs [19]. This finding is in line with previous tiotropium Respimat trials that included an add-on salmeterol arm, and concluded that tiotropium as add-on to ICS was non-inferior to salmeterol as add-on to ICS [9,10]. ACQ-7 responder rates are reported as an indicator of asthma control (eg minimisation of symptoms, activity limitation, bronchoconstriction and rescue β 2 -agonist use); however, the ACQ-7 results presented here should also be interpreted with caution as the study was not designed to show statistical superiority in ACQ-7. Lung function improvements appeared to translate into improved asthma control in favour of tiotropium Respimat 5 μg over placebo Respimat at Week 24, but by Week 52 ACQ-7 responder rates were similar across all treatment groups. The improvements in ACQ-7 with the addition of tiotropium Respimat are even more striking considering that the majority of patients (almost 60%), including those in the placebo Respimat group, were receiving background ICS and LABA and the remainder at least ICS. Furthermore, placebo responses are not uncommon in asthma trials using ICS as background, likely attributable to improved adherence to ICS in clinical trial settings [20]. An earlier pooled analysis from two Phase III studies (NCT01172808 and NCT01172821) comparing tiotropium Respimat 5 μg or 2.5 μg or placebo Respimat also showed that, at 24 weeks, tiotropium Respimat 5 μg and 2.5 μg significantly improved the ACQ-7 responder rate compared with placebo [21]. Therefore, the totality of evidence suggests that tiotropium Respimat 5 μg use may have beneficial consequences for overall asthma control. The efficacy of tiotropium Respimat 5 μg was greater than that of tiotropium Respimat 2.5 μg; the treatment difference for the 2.5 μg dose versus placebo Respimat peaked at Week 12 for trough FEV 1 response and at Week 24 for in-clinic trough PEFR response, but did not clearly separate from placebo for either end point. The efficacy of the 5 μg dose has been demonstrated by previous studies [7][8][9], two of which enrolled Japanese patients [8], and the data presented here provide further support for the benefits of this dose. Compliance with the treatment regimen was high (>90%), indicating that the trial results are a reliable portrayal of the efficacy of tiotropium Respimat. Conclusions In this trial, the tolerability profile of once-daily tiotropium Respimat was confirmed in patients from Japan who had moderate to severe symptomatic asthma while receiving mediumdose ICS, with or without a LABA, as maintenance therapy. The safety and efficacy profile of tiotropium Respimat reported in this trial was in line with previous studies and should be broadly applicable to other patient populations with symptomatic asthma. Over 52 weeks, AEs were generally mild and non-serious, and the overall tolerability profile of tiotropium Respimat was comparable with that of placebo Respimat. Lung function and symptom control data indicate that tiotropium Respimat shows sustained efficacy, with significant improvements over placebo Respimat in trough FEV 1 and trough PEFR parameters over 1 year. Supporting Information S1 CONSORT Checklist. (DOCX) S1 Fig. Trough FVC response for patients receiving tiotropium Respimat add-on or placebo Respimat add-on. Trough FVC response for patients receiving tiotropium Respimat or placebo Respimat as add-on to maintenance therapy of inhaled corticosteroids, with or without long-acting β 2 -agonist, over the 52-week study period (full analysis set). Tiotropium Respimat and placebo Respimat dosed once-daily in the evening. FVC, forced vital capacity. (TIF) S1 File. Trough forced vital capacity response. (DOCX) S1 Protocol. (PDF) S1	2016-05-12T22:15:10.714Z	2015-04-20T00:00:00.000	{ "year": 2015, "sha1": "3c87be90a70b45a763d49e57e14058b2c56611a8", "oa_license": "CCBY", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0124109&type=printable", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "3c87be90a70b45a763d49e57e14058b2c56611a8", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
219033097	pes2o/s2orc	v3-fos-license	Clinical relevance of lipid panel and aminotransferases in the context of hepatic steatosis and fibrosis as measured by transient elastography (FibroScan®) Background Nonalcoholic fatty liver disease (NAFLD) is one of the most common causes of chronic liver disease and is associated with various co-morbidities. Transient elastography (FibroScan®) is a non-invasive method to detect NAFLD using the controlled attenuation parameter (CAP). We aimed to evaluate the association of the lipid panel and aminotransferases concentrations with the presence or absence of steatosis and fibrosis. Methods One hundred and five patients with NAFLD were included. Hepatic steatosis was quantified by CAP (dB/m) and liver stiffness by Kilopascals (kPa), these values were then analyzed against patient lipid panel and serum concentrations of the liver enzymes aspartate aminotransferase (AST) and alanine aminotransferase (ALT). A correlation and multiple regression were used. Mann-Whitney U test was used as non-parametric analysis. Results We observed an association between hepatic steatosis and total cholesterol (B = 0.021, p = 0.038, Exp (B) = 1.021, I.C = 1.001-1.041) as well as serum triglycerides (B = 0.017, p = 0.006, Exp (B) = 1.018 and I.C = 1.005-1.030). Similarly, we found an association between significant hepatic fibrosis and lower concentrations of total cholesterol (B = -0.019, p = 0.005, Exp (B) = 0.982 I.C = 0.969-0.995) and elevated AST (B = 0.042, p = 3.25 × 10-4, Exp (B) = 1.043 I.C = 1.019-1.068) independent of age, gender and BMI. Conclusions Our results suggest that, total cholesterol and triglyceride concentrations positively correlate with hepatic steatosis while significant hepatic fibrosis is associated with lower total cholesterol and higher AST concentrations. Introduction Non-alcoholic fatty liver disease (NAFLD) is one of the most common causes of chronic liver disease worldwide with a global prevalence of 25%. In NAFLD, isolated steatosis can progress to advanced stages with non-alcoholic steatohepatitis (NASH) and fibrosis, increasing the risk of cirrhosis and hepatocellular carcinoma (1). NAFLD frequently coexists with various diseases such as type 2 diabetes mellitus (>70% of patients), obesity (30 -100% of patients) and cardiovascular diseases (20-92% of patients) (1-3). The liver plays a key role in lipid metabolism. It functions in both the uptake and degradation of chylomicrons as well as in the synthesis of lipoproteins and in lipogenesis. In NAFLD an imbalance occurs between the uptake, secretion and synthesis of lipids in the liver which causes an excessive accumulation of macro and/or microvesicles rich in triglycerides in hepatocytes, reaching at least 5% of the total liver weight (4,5). Although biopsy is the gold standard for the evaluation of hepatic steatosis and fibrosis, it is an invasive procedure that, in a minority of cases, can lead to serious complications. Transient elastography (FibroScan®) is a recently developed, non-invasive, sensitive, and specific method that uses certain physical and ultrasonographic parameters to quantify steatosis and liver fibrosis (6)(7)(8)(9)(10)(11)(12)(13). Several studies have demonstrated the usefulness of the lipid panel as an indirect marker of chronic liver disease (14)(15)(16)(17). However, there is little evidence showing the association of serum lipid concentrations and aminotransferases with certain non-invasive quantitative parameters of hepatic steatosis and fibrosis, such as transient elastography (FibroScan ® ) with controlled attenuation parameter. The present study aims to evaluate the association of the lipid panel and aminotransferases with the values of FibroScan ® of NAFLD. Patients This was a cross-sectional study including 105 patients over 18 years of age with NAFLD of varying severity covering the whole spectrum of disease (Isolated steatosis and liver fibrosis with/without steatosis), who were treated at the Gastrointestinal and Liver Speciality Clinic at Merida, Yucatan, Mexico between January 2014 and May 2019. The demographic characteristics, anthropometric measures, and laboratory tests such as serum triglyceride concentration, total cholesterol, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), aspartate aminotransferase (AST) and alanine aminotransferase (ALT), were obtained with the patient's consent and they were measured according to the manufacturer's instructions with standardized methods. Exclusion criteria included significant alcohol intake (≥20 g/day for women and ≥40 g/day for men), clinical evidence of viral or autoimmune hepatitis, drug-induced fatty liver, or other metabolic liver diseases (such as haemochromatosis or Wilson's disease). Also, patients who were being treated with lipid lowering drugs or that declined participation in the study were excluded. This study was designed and conducted according to the principles of the Declaration of Helsinki and was approved by a Local Institutional Ethics Committee (MF-3002 Research Committee Hospital Hispano, Guadalajara, Mexico). Informed consent was obtained from each participant. Measurement of hepatic steatosis and fibrosis The measurement of hepatic steatosis was performed through the CAP expressed in decibels per meter (dB/m) and the determination of fibrosis was performed through liver stiffness expressed in kPa, both measurements were made through Fibro Scan ® transient elastography (FibroScan 501, Echo sens, Paris, France). The FibroScan ® is a non-invasive method with high sensitivity and specificity for determining liver stiffness and steatosis and it has been previously described and validated in populations with 1.005-1.030). Similarly, we found an association between significant hepatic fibrosis and lower concentrations of total cholesterol (B = -0.019, p = 0.005, Exp (B) = 0.982 I.C = 0.969-0.995) and elevated AST (B = 0.042, p = 3.25 × 10 -4 , Exp (B) = 1.043 I.C = 1.019-1.068) independent of age, gender and BMI. Conclusions: Our results suggest that, total cholesterol and triglyceride concentrations positively correlate with hepatic steatosis while significant hepatic fibrosis is associated with lower total cholesterol and higher AST concentrations. Klju~ne re~i: lipidi, enzimi jetre, steatoza, fibroza NAFLD (6). Measurements were performed by placing the tip of the transducer of the FibroScan ® over the right lobe of the liver within intercostal spaces (typically between the ninth and eleventh intercostal space, at the mid-axillary line), with the patient lying in dorsal decubitus with the right arm in maximal abduction. This procedure was performed by expert personnel (>500 measurements) and reliability of the study was defined as ratio of the interquartile range (IQR)/median of liver stiffness measurement reflecting the variability of the total valid measures and should it not exceed 25%. Only examinations with at least 10 valid individual measurements were deemed valid. The presence of hepatic steatosis was defined by a CAP value ≥180 dB/m. Hepatic fibrosis was categorized as non-significant when liver stiffness was <7.9 kPa and significant when it was ≥7.9 kPa (7,9,12,13). Statistical analysis Data were reported as mean ± standard deviation for normal variables and median with interquartile range (25 and 75 percentile) for non-normal continuous quantitative variables. The distribution of the variables was analyzed using the Kolmogorov-Smirnov test. For statistical analysis in quantitative variables, the Student's t or Mann-Whitney U tests, according to data normality were used. The association between hepatic steatosis and liver fibrosis with lipid panel was assessed through univariate linear regression analysis. A multiple linear regression analysis was performed considering the variables with indi-vidual significance of the correlation analysis, where age, gender and body mass index (BMI) were considered covariates. The results were considered significant with a p value <0.05. The statistical program SPSS (IBM SPSS Statistics for Windows, version 25.0, Armonk, NY, USA) and GraphPad Prism (version 6.0; GraphPad Software, Inc., La Jolla, CA) for graphics, were used. Patient characteristics A total of 105 patients were included, of which 59 (56%) were men. The mean age was 52.7±12.1 years old. Hepatic steatosis was present in 83 (79%) patients. According to liver stiffness, 35 (33.3%) patients showed significant hepatic fibrosis. The anthropometric, clinical and biochemical characteristics are shown in Table I. Total cholesterol (p=0.0001), LDL-C (p=0.0072) and triglycerides (p=1.9x10 -5 ) serum concentrations were significantly higher in patients with hepatic steatosis. However, steatosis was not associated with significant differences in aminotransferase values. Regarding liver stiffness, patients with significant hepatic fibrosis showed significantly lesser serum concentrations of total cholesterol (p=0.0126), LDL-C (p=0.0405) and triglycerides (p=0.0064) than those with non-significant hepatic fibrosis. AST values (p=0.0006) were significantly higher in patients with significant hepatic fibrosis. Boxplot graphic of medians with interquartile range are shown in Figure 1. The results are expressed as mean ± SD and median (25 th and 75 th percentile); BMI, body mass index; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol. Table II). Similarly, an association between significant hepatic fibrosis and concentrations of total cholesterol, LDL-C, triglycerides and AST was also evaluated. The results showed that the presence of significant hepatic fibrosis was associated with decreased (Table III). Discussion In clinical practice, the diagnosis of NAFLD is typically established through the combination of laboratory and radiographic findings with liver biopsy being the gold standard. However, these methods have limitations and provide a qualitative picture of steatosis and fibrosis of the liver that limit the reproducibility and practicality of the diagnosis (4). Recently, non-invasive methods have been described for the quantification and surveillance of steatosis and stiffness such as the CAP and kPa of FibroScan ® , which have demonstrated a strong correlation with the accumulation of fat and fibrosis in the liver (7,10,12,13). The results of our study showed that total cholesterol, triglycerides and serum LDL-C concentration were significantly higher in patients with hepatic steatosis. Observational studies such as DeFilippis et al. (18) and Loria et al.(19), showed that patients with NAFLD had a lipid profile consistent with an atherogenic phenotype showing dose-dependent characteristics such that the more severe their hypertriglyceridemia, the more severe their hepatic steatosis. The pathogenesis of NAFLD is one of insulin resistance, an imbalance in the metabolism of carbohydrates and lipids, and lipotoxicity that leads to hepatocellular damage, inflammation and fibrosis (14,15). The accumulation of fat in the form of lipid droplets in liver tissue is a fundamental histopathological characteristic for the diagnosis of NAFLD. Most commonly these lipid droplets are composed mainly of triglycerides (16,17,20). Moreover, diet patterns also main role in NAFLD independently hyperlipidemia, diabetes and obesity. Musso et al. (21) found that subjects with NAFLD consumed a diet richer in saturated fatty acids and poorer in polyunsaturated fatty acids, fiber, and antioxidant vitamins C and E compared with controls. In addition, Yasutake et al. (22,23) found that non-obese patients with NAFLD had significantly greater intake of dietary cholesterol and significantly lower intake of polyunsaturated fatty acids compared with an obese group with NAFLD. The main feature of dyslipidemia in patients with NAFLD is a normal LDL-C atherogenic lipid profile that consists of high levels of triglycerides, low levels of HDL-C and an increase in small and dense LDL particles. However, in this study there was a correlation observed between hepatic steatosis and LDL-C. Despite was a correlation observed between hepatic steatosis and LDL-C, this not was associated after a multiple linear regression analysis, perhaps LDL-C underestimates the true cholesterol load in NAFLD since their concentrations do not completely capture the total mass of lipoprotein particles (24). On the other hand, we observed a significantly lower concentration of total serum cholesterol, triglycerides, and LDL-C in patients with significant hepatic fibrosis. After adjusting for age, gender and BMI, only total cholesterol and AST values remained significant. These findings can be explained by a decline in the liver's ability to capture, synthesize and mobilize lipoproteins and triglycerides due to the loss of hepatocytes in patients with significant hepatic fibrosis (25). Similar results have been previously described by Tietge et al. (26) in a cohort of 52 patients with cirrhotic livers and 16 patients with clinically stable liver transplants. They found that patients with cirrhosis had a reduction in the clearance of free fatty acids and a diminished ability to synthesize cholesterol and triglycerides when compared to transplant patients, indicating the loss of primary liver function. In contrast to the lipid panel, AST and ALT are commonly used as indicators of hepatocellular injury (27). The evidence has shown than elevations in AST confer an increased risk of all-cause mortality (28). In NAFLD, AST elevations are frequent, but they are more pronounced in the presence of significant hepatic fibrosis and cirrhosis (27). Similar results were found in our study. Foremost among the limitations is that due to the transverse nature of cross-sectional studies like ours, the findings are correlational only in nature and do not establish causality. In conclusion, our results suggest that total serum cholesterol and triglyceride concentrations are positively correlated with hepatic steatosis while significant hepatic fibrosis is associated with lower total cholesterol and higher AST concentrations in patients with NAFLD according to FibroScan ® results. These findings suggest that the lipid panel and aminotransferases studies may serve as simple and useful metabolic markers for the identification and follow-up surveillance of those individuals at risk for NAFLD. Large prospective cohort studies are still needed in order to validate and confirm our results. Acknowledgements. To Jonathan Patrick Alessi for the critical analysis and the English revison. Eloiza Narváez for data organization and Hepatoescaner del sureste SCP for the facilitation of ethical access to FibroScan ® data. Funding The author received no financial support for the research. Conflict of interest statement The authors declare that they have no conflicts of interest in this work.	2020-04-30T09:08:17.262Z	2020-04-28T00:00:00.000	{ "year": 2021, "sha1": "7bb3ca184c14dcf031eb7c967b318a967b9c493c", "oa_license": "CCBY", "oa_url": "https://scindeks-clanci.ceon.rs/data/pdf/1452-8258/2021/1452-82582101060C.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "34065c9ae0bd8d7385a3226f6c14e4d89d726ca3", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
255259517	pes2o/s2orc	v3-fos-license	MACROMEDIA FLASH 8 LEARNING MEDIA TO INCREASE LEARNING INTEREST IN THEME 2 STUDENTS ALWAYS SAVE ENERGY CLASS IV SD NEGERI 016555 ASAHAN , Introduction The development of the world of education is very significant in line with the development of science and technology. The world of education is always expected to follow in the footsteps of global technological developments, this is a demand, because education is the main capital in building the younger generation, educating the nation's life, and preparing to become a reliable and competitive workforce. The world is developing, technology is developing and the world of education is growing. The world of education often depends on technology, because it can help in improving learning and facilities and infrastructure that are increasingly needed in the world of education to uphold the quality of education (Rahmi et al., 2019) . Meaningful and effective learning will take place if it gives satisfactory results for all parties such as teachers and also the students themselves. Teachers will be satisfied if students get optimal learning outcomes, and can develop the potential of students (Hidayah, 2018) .Interest can be obtained through learning, because by learning students who initially do not like a particular lesson, with increasing knowledge about the subject, interest grows so that students will be more active in learning the lesson. Interest is a constant desire to pay attention or do something, interest can create enthusiasm in carrying out activities so that the goals of these activities can be achieved, and the existing spirit is the main capital for each individual, especially students in carrying out teaching and learning activities (Dyah Anungrat Herzamzam, 2018) . (Kustandi, Sutjipto, 2019) Learning Media is a tool that can help the teaching and learning process and serves to clarify the meaning of the message conveyed, so that it can achieve learning objectives better and more perfectly. Ariani and Dany (2018) say that "multimedia is the result of a combination of various media in the form of text, images, graphics, sound, animation, and video that are used to convey messages to the public". Macromedia Flash 8 is a multimedia that can create videos, animations, images, and sounds in an easy and effective way. By using multimedia, abstract things can be concreted so that they can be displayed in front of students and attract interest in learning through various forms of animation that are presented (Fakhri, 2018). The use of media is able to make the teaching and learning process more practical and efficient. (Helsa et al, 2019) stated that multimedia can make it easier for students to understand learning and apply the knowledge gained from the learning process into thematic problems. Thus I choose multimedia-based learning media, namely Macromedia Flash 8 to increase student interest in learning, besides that in today's technology has been widely used in education, so that by using this multimedia-based learning media, students and educators are not left behind with technological developments. which will be applied during the learning process (Natasya & Izzati, 2020) . According to (Sukmawari et al, 2022: 202) learning is needed in order to prepare student face era revolution industry 4.0 which demand Skills century 21, that is think creative, think critical, communicating and collaborate. (Rangkuti & Sukmawati, 2022). A good learning process begins with wise planning. In study students not only interact with the teacher but , students as well interact with learning resources used to achieve good learning outcomes desired. (Hidayat and Khayroiyah: 2018) to reduce the emergence of obstacle study, so teacher need prepare device learning which appropriate. Innovations learning which demand power educator nor participant educate for think creative as well as capable adapt with the times to produce active, creative, innovative students and of course have noble character (Sukmawarti et al., 2021). According to in this modern era, technology is developing in various fields, such as education, including at the basic education level. Method Study It uses Research & Development (R&D) development. According to Sugiyono (2017) research and development (R&D) methods are research methods used to produce certain products, and test the effectiveness of these products. In this study, an audio-visual learning media was developed which was designed through the Macromedia Flas 8 application on the theme 2 Always Save Energy. Place, Time and Research Subject The research was conducted at SD Negeri 016555 Asahan, which is located in Bandar Pulau sub-district, Asahan district. This research was conducted in the academic year 2022/2023. Determination of research schedule in September 2022 . The subject in this research is learning media Macromedia Flash 8 . Data Collection Instruments and Techniques The instrument is a measuring tool used in research to obtain valid data. The data from this study were obtained through questionnaires or questionnaires, observation and documentation. The instrument used in this research is to measure the feasibility of the product developed by the researcher. Data analysis technique From the data that has been obtained from the feasibility aspect of the resulting product. The data analysis technique used in this research is media feasibility analysis. Data analysis uses quantitative data to test the feasibility of the product being developed which is taken from the results of teacher and student response questionnaires using a Likert scale. (Riduwan dalam Zuhriyah, 2019:485) 3. Result and Discussion Research and development of learning media Macromedia Flash 8 application in thematic learning always saves energy for fourth grade elementary school. Researchers used the ADDIE model . To design a learning system, which consists of 5 stages, namely; Analysis Phase (Analysis) At this stage, what is done is 1) needs analysis : At this stage the needs analysis aims to find out how to use learning media at SD Negeri 016555 Asahan , 2) Analysis of Media Expert Validation The results of the validation of Sub-theme 1 learning media Macromedia Flash 8 Application are categorized as "Very Eligible" with a total score of 52 out of a maximum score of 60, so the percentage of eligibility is 86.6% with the final conclusion from the media expert validator that the learning media is feasible without revision so that researchers can proceed to the next stage. The results of the validation of Sub Theme 2 learning media Macromedia Flash 8 Application are categorized as "Very Eligible" with a total score of 50 from a maximum score of 60, so the percentage of eligibility is 83.3% with the final conclusion from the media expert validator that the learning media is feasible without revision so that researchers can proceed to the next stage. The results of the validation of Sub-theme 3 learning media Macromedia Flash 8 Application are categorized as "Very Eligible" with a total score of 52 out of a maximum score of 60, so the percentage of eligibility is 86.6% with the final conclusion from the media expert validator that the learning media is feasible without revision so that researchers can Educational Practitioner Validation The learning design validation (RPP) was carried out by Mr. Sujarwo, S.Pd., M.Pd as a learning design validation carried out on October 4, 2022. There are 6 aspects, namely the identity of the lesson plans, aspects of the formulation of learning objectives and indicators, material aspects, aspects of selection learning approach, aspects of planning learning activities, aspects of learning resources, and aspects of language with a Likert scale of 1-5. From the results of the validation of education practitioners in sub-theme 1 by FKIP lecturers, they obtained a total score of 47 out of a maximum score of 50 with a percentage of 94% and was included in the "Very Eligible" category. From the results of the validation of education practitioners in sub-theme 2 by FKIP lecturers, they obtained a total score of 47 out of a maximum score of 50 with a percentage of 94% and was included in the "Very Eligible" category. From the results of the validation of education practitioners in sub-theme 3 by FKIP lecturers, they obtained a total score of 47 out of a maximum score of 50 with a percentage of 94% and included in the "Very Eligible" category. So it can be concluded that the learning media of the Macromedia Flash 8 Application in the Thematic Learning Theme Always Save Energy in Grade IV Elementary School is determined to be feasible and ready to be used for Grade IV Elementary School. Material Expert Validation Material validation was carried out by Mrs. Nurasiyah S.Pd who is an elementary school teacher 016555 Asahan. Validation by material experts aims to obtain information, criticism, and suggestions so that the learning media of the Macromedia This implementation phase aims to determine the feasibility of implementing Macromedia Flash 8 Application media on the Thematic Learning of Always Saving Energy in Class IV Elementary School which will be tested on students. By doing at the implementation stage, the researcher implements it in stages to see whether the media developed is valid or not and therefore the researcher conducts individual trials and only small group trials. Big Indonesian Dictionary (KBBI) and no. item 10 contains the completeness of the information submitted is in accordance with the material. The results of the percentage of eligibility that have been obtained prove that there is a positive response from all respondents or students using the learning media of the Macromedia Flash 8 application that was developed, so that it can be categorized as "Very Eligible". Results of the Evaluation Phase (Evaluation) The last stage in the ADDIE development model is the evaluation stage. At this evaluation stage, it is carried out to see the feasibility, practicality, and effectiveness of each product assessment process that has been carried out at the time of validation. Based on the score results from the feasibility of learning media Macromedia Flash 8 Application which has been validated by media experts in sub-theme 1 with a score of 86.6%, sub-theme 2 with a score of 83.3%, sub-theme 3 with a score of 86.6%, and experts the material in sub -theme 1 gets a score of 94 %, sub-theme 2 gets a score of 90 %, sub -theme 3 gets a score of 92%, and the responses of education practitioners in sub-theme 1 get a score of 94%, subtheme 2 gets a score of 94%, sub-theme 3 get a score of 94%, on the learning	2022-12-30T16:13:25.912Z	2022-10-14T00:00:00.000	{ "year": 2022, "sha1": "2c780e7fa9741841809a6f2f045a83a19b4d08c0", "oa_license": "CCBYSA", "oa_url": "https://journal.trunojoyo.ac.id/widyagogik/article/download/17211/7233", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "755ae1b68e2f5ee481a4632394b7776f76cb8bcd", "s2fieldsofstudy": [ "Education", "Computer Science" ], "extfieldsofstudy": [] }
258296636	pes2o/s2orc	v3-fos-license	Bursaries Reimagined: Addressing Digital Inequity through a Library-Led, University-Wide Laptop Bursary Program Abstract The rapid switch to online learning in early 2020 exacerbated problems students were already having with obtaining and maintaining up-to-date devices and a reliable internet connection. MacEwan University Library began offering 4-month term laptop loans at the beginning of the pandemic, but it was clear this was not fully meeting student needs. In response to conversations with faculty and students, the library secured funds from the university’s Student Technology Fee to launch a laptop bursary pilot in Winter 2022, which in turn expanded to a university-wide bursary in Fall 2022. This article discusses why an in-kind laptop bursary was the right approach at the right time in this setting; how this initiative contributes to equity and accessibility; and finally, perceptions of the value of this work, its fit within the scope of the library, and how the unique position of the library as a student-focused service and academic unit positioned it well to successfully offer this bursary. Challenges and opportunities for improvement are also discussed. Introduction A digital divide exists amongst our student community. Some have stable, ready access to a laptop and internet while others, to varying degrees, are reliant on campus and other publicly accessible computers and wifi. This divide intensified during pandemic circumstances, disproportionately affecting students already grappling with socioeconomic disparities (Lederer et al. 2021). While students from across the schools and faculties at MacEwan University encountered these challenges, this lack of digital equity emerged as a leading concern for the university's pimâcihisowin Foundation Program 1 (pimâcihisowin) and Child and Youth Care program 2 (CYC). pimâcihisowin and Child and Youth Care draw students from communities historically underrepresented in postsecondary institutions, and are relatively small, close-knit programs at MacEwan. pimâcihisowin, in nêhiyawêwin (the Cree language), means "to create a life of independence." Launched in 2019, it helps Indigenous students bridge possible gaps between high school credits and the entry requirements for post-secondary diplomas and degrees. It offers students an opportunity to learn and unlearn together as they explore and apply Indigenous knowledge in their development of a student identity as individuals and as participants in western academia. Sustained relationality is paramount to learner success in pimâcihisowin. Reliable technology and connectivity support this togetherness and success, and supports student advancement into diploma and degree programs. Child and Youth Care, in turn, was one of the earliest degree programs offered at MacEwan. It attracts students from an array of communities and circumstances and holds equity admission spaces in particular for students identifying as Indigenous (First Nations, M etis, or Inuit Peoples in Canada). As a highly relational profession, the CYC curriculum and methods of instruction are seeded with collaborative and experiential coursework and assignments. Additionally, there is faculty cross-teaching between pimâcihisowin and Child and Youth Care, creating synergy between the two programs. As instruction shifted online, faculty and students in these programs shared an acute awareness that a sizable number of students no longer had the means to fully participate in online classrooms and coursework. They also brought these concerns into conversations with their subject librarian. One of the biggest challenges was access to reliable and consistent technology to complete assignments, attend online lectures, and to develop their researching skills. While these challenges existed prior to the pandemic, they were aggravated by isolation measures mandated by the provincial government and put in place by the university to support campus and broader community health. For some students, stable internet access was disrupted by a return home to rural communities with unreliable network coverage. Other students had to decide between basic living expenses and technology tools, or between personal and work obligations and extraordinary efforts to access campus and/or public technology tools as required for academic success. Sharing these experiences with their librarian, the faculty and students were looking for solutions, possibly through the library's technology loans. Similar to other libraries, such as the University of Alabama (Decker, 2021), MacEwan Library and campus partners worked during the pandemic to create term loans for laptops and wifi sticks. However, it was clear that this was not enough: these students needed ongoing, stable access to computers and connectivity. Laptop bursary pilot winter 2022 In response to the program feedback, MacEwan Library successfully submitted a proposal to the Student Technology Fee Fund, and in partnership with Technology Support and the Office of Financial Aid, piloted a laptop bursary in Winter 2022. In line with other bursaries at the university, a financial need threshold was set. The bursary pilot was also limited to students enrolled in either pimâcihisowin or Child and Youth Care in Winter 2022 (184 students in total). Faculty and staff connected to the two programs and their subject librarian worked together to advertise the pilot and extensively assist interested students with their bursary applications. Of twenty-four applications, twenty met the bursary criteria. One student had managed to secure a device elsewhere, and so nineteen students were ultimately awarded a brand-new laptop and offered access to paid internet for a year via wifi sticks. Based on anecdotal student and faculty feedback collected via survey and in-person conversations, common reported impacts of receiving a laptop were: An increased sense of investment by the university in the students and their programs An increased sense of belonging by students A sense of relief from access and technology issues that were causing financial and relational stress and hindering academic focus and engagement An increased sense of confidence in launching careers and pursuing further study opportunities. With respect to receipt of laptops, the majority were picked up by students within 48 hours of notification. However, a few laptops were picked up late in term, weeks after their awarding. This served as a reminder of the complexity of individual circumstances. Obstacles exist for students, and they exist despite best intentions and efforts. Possible reasons that emerged through stakeholder conversations included mixed feelings around receipt of a device; an unfamiliar pick-up space; as well as timing and convenience factors. Having worked with these students on their initial applications, however, the subject librarian encouraged and facilitated the pick-ups. Expansion to university wide laptop bursary distribution fall 2022 After the success of the pilot, the laptop bursary was increased with the help of additional funds from the student technology fee and from the Office of the Provost project fund. A total of 101 laptops were purchased, and the bursary was opened to all students at MacEwan University in Fall 2022. The financial aid office at MacEwan University regularly receives 600-800 bursary applications per academic year. A total of 619 applications to the laptop bursary were received, and of these, 491 met the bursary criteria. The method of awarding was the same as the pilot wherein students with the greatest demonstrated financial need received the bursary. All 101 laptops were awarded. Similar themes emerged in faculty and student feedback across both deliveries of the laptop bursary. The expansion of the program, however, resulted in additional anecdotal feedback from university staff. In particular, staff in the library, makerspace, writing center, and access and disability services shared feedback not only about student technology needs amongst the students they support, but also of student excitement upon being told about the bursary. Moreover, library staffcoping with waves of student requests for laptops during periods where all short and long-term laptops were in usewere happy to be able to direct students to apply for the bursary. The expansion of the program led to a fairly even distribution of laptops to students across all university programs. In the process, however, the number of laptop recipients enrolled in pimâcihisowin and CYCW dropped. Under the current iteration of the bursary, successful students not only had to meet the general financial threshold for bursaries at MacEwan University, but also had to be within the 101 spots assessed as being in most financial need. While analysis of the financial assessment process determining these spots is beyond the scope of this article, it should be acknowledged that a brief snapshot of an individual's financial circumstances at the time of application can miss important context around equity and accessibility in the postsecondary environment. Additionally, there is no guarantee of the students most in need applying for the bursary. The complexity and invasiveness of applying for bursaries in general is a sizable investment in time, in effort, and in emotions for students and their helpers. Such concerns were shared by applicants, and echoed by helpers at kihêw waciston, the Indigenous Student Center at the university. Finally, the application process itself required stable access to the very technology tools being sought in the first place. Ultimately, the laptop bursary initiative put Student Technology Fees back into the hands of students who need it most to support their academic success. It is a novel way to welcome our first generation, low income, and/or minority students into post-secondary life. If continued, this bursary can be an important piece in promoting equity and diversity at MacEwan University. Future consideration will also be given to reserving a percentage of the laptops for Indigenous students in particular. Technology access and equity in postsecondary Koch (2022) speaks to a complex digital divide in Canada that exists not only in rural areas and in Indigenous communities, but in urban settings as well-typically the site for postsecondary institutions. She notes a role for all levels of government in addressing this, and also points to stakeholders such as education ministries and institutions as mechanisms for change (2022, p. 96). To date, however, there are limited options for stable device access and connectivity for students through their postsecondary institutions in Canada. With respect to student devices, the Canadian post-secondary context is markedly different than the American one. In the United States (US), there are several programs that support the provision of devices to all incoming students at many participating universities (Miller, 2022). An example is the Wireless Mobile Computing Initiative, launched in 2004, through which, Dakota State University offers a free device to every first-year student (Dakota State University, 2020). In Canada, however, it is rare for such broad technology initiatives to be in place at postsecondary institutions. We conducted a canvas of Canadian university, college and technical institute websites in Fall 2022 and did not find evidence of any programs currently providing laptops. In Canada, it is more common for laptop specifications or requirements to be stated in general or for specific programs (e.g. University of Calgary 3 , MacEwan University 4 , Southern Alberta Institute of Technology 5 (SAIT), and University of Toronto 6 ). Several retailers and technology companies offer education discounts on laptops (e.g. Best Buy 7 , Apple 8 and Lenovo 9 ), and sometimes students have access to technology discounts from campus bookstores (e.g. University of Lethbridge 10 ). University of British Columbia 11 also offers laptop and internet discounts to remote students. However, for the most part, students are expected to come into their studies with reliable technology that meets specified or unspoken expectations around coursework needs. Similar to peer institutions, MacEwan University has no broad technology provision program in place. It appears that MacEwan University is unique in its creation of an in-kind technology bursary, for we have not found a similar bursary presently offered by another Canadian university. There are real consequences for academic achievement when students cannot obtain and maintain technology. Building on previous literature, Gonzales et al. (2020) surveyed 748 undergraduate students and conducted three focus groups at a midwestern university to explore connections between access to technology and benefits for student academic achievement along with connections between the lack of sustained technology access and student grades. Students shared how technology and connectivity issues have resulted in them missing important information about assignments; have required them to seek out assignment extensions; and have left them unable to submit assignments altogether in some cases. Key findings were that students of lower socioeconomic position disproportionately experienced problems with maintaining access to technology, software, and connectivity, and that even when adjusting for socioeconomic and other factors, students with "more poorly functioning laptops and laptops that broke down more often also had lower GPAs" (p. 760). Reisdorf et al. (2020) suggested that universities should identify and remedy non-ownership of devices as early as possible to prevent predictable student achievement difficulties. They noted that in addition to device ownership, device maintenance affects student achievement, too, and this challenge to ensure the working conditions of devices across a program of study warrants further investigation. Adding in the risk of cumulative deficits with digital literacies, which can further disadvantage student academic achievement as they go along in their studies (van Deursen & van Dijk, 2019), it is advisable that postsecondary institutions adopt interventions addressing device ownership, maintenance, and utilization (Reisdorf et al., 2020). Toward addressing the pitfalls of maintenance, MacEwan University's Laptop Bursary provides a warranty that students can access through the laptop vendor, and university technology support helps recipients with setting up the laptop, installing software, and connecting on campus. Shank and Cotten (2014) found that youth development of self and academic efficacy were positively associated with variables including daily access to a computer, the amount of time spent using it, and the activities students engaged in while using the computers. When considering what is most advantageous for students in terms of ownership versus borrowing a library laptop, there are some notable tradeoffs that may adversely affect these specified variables. For ownership, students are responsible for maintenance, such as repairs and upgrades, and this comes with costs and challenges (Gonzales et al., 2020;Arch & Gilman, 2019). Still, ownership of a functioning laptop does create the condition of consistent individual access, which, in turn, creates circumstances where students can engage in the activities required to develop digital literacy and fully participate in their academics. Borrowing a library laptop, on the other hand, offloads maintenance to the lender but comes with a deadline after which access is no longer assured. Additionally, lenders may mandate deep freeze software onto loaner laptops that complicate student efforts to download and use software outside of the standard institutional image. This limits the likelihood of students developing certain digital literacy skills. Perceptions of value, library scope, and positioning The critical connection between libraries and access is clearly stated in the American Library Association's statement of core values of librarianship, such that information should be "equitably accessible to all library users" (2019). The importance of the role that technology plays in the modern library's commitment to providing access to information as part of global sustainability is also well-represented by the International Federation of Library Associations and Institutions' (IFLA) report, Access and Opportunity for All: How libraries contribute to the United Nations 2030 Agenda. This 2016 report states, " … increasing access to information and knowledge across society, assisted by the availability of information and communications technologies (ICTs), supports sustainable development and improves people's lives" (p.1). The value libraries place on providing laptops to students is evidenced by the many academic libraries that have established laptop lending programs. The proliferation of these programs recognizes the importance of addressing this first hurdle of the digital divide. However, it is possible that libraries have become complacent, seeing the provision of technology as sufficient in itself and thus shifting focus to what has been characterized as a second level of the divide, digital literacy (Cohron, 2015;Rowsell et al., 2017). Goedhart et al. (2019), discussing their focus on lower socioeconomically positioned mothers within a large-scale study of the digital divide in the Netherlands, caution that it can be tempting to think that access to devices is a solved issue in well off countries when in fact it is clearly still a "substantive" issue for lower socioeconomic and intersectional populations (p. 2360). Foster et al. (2022) note that in addition to laptop lending programs, there is a need for longer-term strategic investments and solutions for the "seemingly endless" demand (p. 102). It is possible that the additional challenges presented by the pandemic have provided impetus for academic libraries to recommit to creatively addressing this initial hurdle, so crucial to access, which is disproportionately affecting students from lower socioeconomic backgrounds. It may seem self-evident to library workers that providing laptops to students is an extension of the library's mission to support academic work and build their confidence in navigating today's world of information. Growing the laptop bursary has recently been identified as a goal for MacEwan University Library in recent planning discussions. As part of student-centered care, providing technology to bridge the digital divide is part of our commitment to removing access barriers to library services and resources. However, even within libraries, it may be that once resources are available on the internet, with proper accessibility features, they are considered accessible. Are they really, however? If a student does not have access to the internet, or if their device is inoperative within the library's digital domain, and they cannot enter our physical library because we are closed due to pandemic or other physical restrictions, do they have access to our collections and resources? Today, does an academic library's responsibility for providing access to information extend to ensuring that students have the minimum devices and internet access to connect? We see this work as an extension of our library's commitment to providing access, supporting academic work, and improving equity. It is important to note, however, that this answer does not completely address the questions of whose responsibility it is-at an academic institution -to ensure that students have laptops. Similar to questions about financial need assessment, this question is beyond the scope of the article. Rather, we consider some perceptions of how the laptop bursary initiative fits with library work and its value at the broader institution level. Outside of the library, the rationale for why this is library work may be less clear to faculty. Some faculty, like those in the pimâcihisowin and Child and Youth Care programs, for example, saw a clear connection. However, we did field questions from a faculty member who, upon observing the stacks of laptops stored in a librarian's office, wondered why obtaining laptops for students would fall to a librarian? Student feedback upon receiving the laptops indicated that what might seem like a transactional service-the provision of a laptop to a student-has larger consequences. Receiving a laptop can impact critical components of student success, such as a sense of belonging and self-actualization (Shank & Cotten, 2014). These benefits may not be as readily associated with meeting these basic technological needs. Having reliable technology could also encourage a student to take advantage of academic and career opportunities and reap rewards they might not otherwise have been able to. One bursary recipient in their final year of studies noted in their feedback an increase confidence in their ability to pursue diverse employment options. Owning a laptop, they felt, put them in a better position to explore work opportunities in the nonprofit sector where employees may not always be able to rely on employer-supplied technology. The library's creation of this bursary received significant support at MacEwan University beyond the partners that were directly involved with making the bursary happen. The Office of Communications and Marketing helped prepare materials to promote applications across campus. The Provost's Office not only contributed funds, but also supported advertisement and uptake of the bursary through a newsletter article. This is akin to support by the Provost's Office for other more obviously academic library initiatives, such as a library-led grant for faculty to develop Open Educational Resources. The Laptop Bursary was praised as an innovative way to leverage existing funds and processes to provide support to students and work toward equity. MacEwan University Library is situated as both an academic and a service department, this was crucial to arranging this nonstandard bursary program. Bridging academics and administration, librarians and the Dean of the Library had the established relationships needed to work with administrative departments (e.g., Information & Technology, Registrar, Financial Aid) as well as academic departments and the Provost's office. This was instrumental in securing funding, ordering the laptops, getting board approval, advertising, and administering the bursary. Challenges and opportunities When discussing the value of this initiative to the university, we continue to seek sustainable funding to maintain and expand the bursary. The Student Technology Fee Fund is in high demand and not intended to provide ongoing funding. We have been able to find internal funding to maintain the bursary for the 23/24 academic year. As for expansion, the library has been working with members of the university's fund development team to approach potential donors and partners for this project. We are actively trying to secure ongoing internal and external funding and/or in-kind donations to maintain the bursary and hopefully expand it. Even if the program continues, expanded or not, it cannot meet all the demands at our institution in the same ways as a program that distributes a device to every first-year student. Nonetheless, its existence is an important, tangible starting place for improving equity and diversity, helping students in need, and relieving stress on the library's laptop and tablet loaning system. Should the laptop bursary disappear, its loss would be felt across the campus and very likely beyond as we know laptops are often shared. Despite the discouraging possibility that this program may not have longevity, the initiative has opened some doors. A stronger relationship and pathways have been formed between the library and the Financial Aid Office for offering in-kind bursaries. There is increased recognition of and excitement for the possibilities inherent in the relationships and collaborations between subject librarians and programs/departments. Another program in the Faculty of Health and Community Studies, which houses Child and Youth Care, is considering a laptop bursary specific for students participating in their distance program. Additionally, early-stage conversations on campus are starting around support for parents struggling with childcare needs, and whether a bursary or other partnerships could provide some relief. These are some of the hopeful shoots to have grown out of the creative use of existing resources and cross-department collaboration, led by the library and its people, to meet students' needs.	2023-04-24T15:02:39.327Z	2023-04-03T00:00:00.000	{ "year": 2023, "sha1": "6647e19eb3be2178f84d5a695f1d61757d59c532", "oa_license": "CCBYNC", "oa_url": "https://doi.org/10.1080/10572317.2023.2198877", "oa_status": "HYBRID", "pdf_src": "TaylorAndFrancis", "pdf_hash": "602dd1ce28d32b4d86424fa8dc6a6f67d0aad8cd", "s2fieldsofstudy": [], "extfieldsofstudy": [] }
66369363	pes2o/s2orc	v3-fos-license	Unipolar ion emission enhances respiratory protection against fine and ultrafine particles We developed a novel concept that allows to considerably improve the performance of conventionally used filtering-facepiece respirators against fine and ultrafine aerosols including airborne viral and bacterial agents. The concept is based on the continuous emission of unipolar ions. The effect was evaluated through the real-time monitoring of the concentration and size distribution of fine and ultrafine aerosol particles. The measurements were conducted inside and outside of a respiratory mask that was face sealed on a breathing manikin. A commonly used Type N95 respirator and surgical mask were utilized for the tests. The manikin was placed in a 24.3-m3 indoor test chamber and exposed to polydisperse surrogate aerosols simulating viral and bacterial particles with respect to the aerodynamic size. The particle penetration through the mask was found to decrease by one-to-two orders of magnitude as a result of continuous unipolar ion emission in the chamber. The flux of air ions migrated to the breathing zone and imparted electrical charges of the same polarity to the aerosol particles and the respirator filter surface. This created an electrostatic shield along the external surface of the filter, thus enhancing the protection characteristics provided by the respirator. The above performance enhancement effect is crucial for minimizing the infectious risk in the cases when the conventional filtering-facepiece respirators are not able to provide an adequate protection against airborne viruses and bacteria. Introduction The outbreaks of emerging diseases (e.g., SARS) and the threat of bioterrorism have triggered an urgent demand for adequate respiratory protection against bioaerosol agents, including airborne viruses and bacteria. Particular interest has been directed towards increasing the e ciency of existing respiratory protection devices. The respirators di er from one another by their ÿltration e ciency, which is dependent on the ÿlter properties and the particle size. For example, a Type N95 respirator may allow up to 5% penetration in "a worst case scenario," when most-penetrating sodium chloride particles of 0:3 m mass median aerodynamic diameter are drawn through the ÿlter at a ow rate of 85 l min −1 (Federal Register, 1995). The penetration e ciency of larger Mycobacterium tuberculosis bacteria (Mtb, 0:8 m) through a face-sealed N95 respirator at strenuous workload is as low as about 0.5% . The face-sealed ÿlter of a conventional health-care mask, which ensures relatively low pressure drop and consequently good comfort level, allows approximately 15% of airborne Mtb surrogate bacteria to penetrate, thus providing 85% protection against these bacteria (Willeke et al., 1996). If the bacterial concentration in the air is 1000 m −3 , an unprotected individual breathing at 30 l min −1 inhales 1800 microorganisms per hour, whereas the one wearing a perfectly ÿt N95 respirator inhales up to 90 microorganisms per hour. If the infectious dose of a bioaerosol agent of interest is less than 90, the N95 respirator may not provide an adequate respiratory protection once the exposure time exceeds one hour. The use of an improperly ÿt-tested tight-ÿtting respirator may further decrease the respiratory protection level because of the additional particle penetration that occurs through the face-seal leaks (Chen et al., 1990;Chen & Willeke, 1992;Oestenstad, Dillion, & Perkins, 1990a;Oestenstad, Perkins, & Rose, 1990b). Based on the above considerations, it seems very useful if the ÿltration e ciency of existing respirators can be increased while the comfort level provided by these devices would remain the same. With respect to the respiratory exposure and protection, the particle aerodynamic diameter range of d a ∼ 0:04-2 m is of special public interest because of its health relevance. Many bioaerosol agents, including viruses and bacteria that cause emerging diseases as well as those that can be used for biological warfare or in the event of bioterrorism, belong to this size range. For example, according to the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov), the dimension of the coronavirus virion (the etiological agent of the SARS) are (60-120) × (160-200) nm, which corresponds to d a ∼ 0:1 m. For Bacillus anthracis (bacteria causing anthrax), d a ∼ 1 m. As the above range is broad and includes both the ÿne and ultraÿne particle fractions (Baron & Willeke, 2001;Hinds, 1999), the particle penetration e ciency through the ÿlter media can be a ected by several mechanisms and is generally characterized by the particle aerodynamic size. This allows testing the performance of respirator ÿlters against pathogenic agents using non-pathogenic aerosol surrogates that simulate the aerodynamic characteristics of the particles of interest. In this study, we developed and evaluated a novel concept that drastically enhances the performance of conventional ÿltering-facepiece respirators against ÿne and ultraÿne aerosol particles. The concept is based on the continuous emission of unipolar ions into the air in the vicinity of the respirator. The aerosol particles are unipolarly charged by air ions primarily due to the di usion charging mechanism (Adachi, Kousaka, & Okuyama, 1985;Frank, Cederfelt, & Martinsson, 2004;Hernandez-Sierra, Alguacil, & Alonso, 2003;Wiedensohler et al., 1994). The ion-ÿlter interaction and the deposition of unipolarly charged particles on the external surface of the ÿlter impose signiÿcant unipolar charge on the ÿlter. This creates a shield for the incoming particles (as they carry charges of the same polarity), which decreases the penetration e ciency through the ÿlter. Experimental measurements The new concept was experimentally evaluated in a non-occupied, unventilated indoor test chamber (L × W × H = 3:78 m × 2:44 m × 2:64 m = 24:3 m 3 ). This facility, developed in the Center for Health-Related Aerosol Studies at the University of Cincinnati, has been used in our previous studies (Choe et al., 2000;Grinshpun et al., 2002Grinshpun et al., , 2004. The experimental setup is schematically shown in Fig. 1. A breathing manikin with a face-sealed respiratory mask was exposed to the airborne polydisperse surrogate aerosols that simulated viral and bacterial particles with respect to their aerodynamic size. The leakage tests were conducted between the mask and the face of the manikin with a bubble-producing liquid (Trubble Bubble, New Jersey Meter Co., Paterson, NJ, USA). The manikin operated at a breathing ow rate, 30 l min −1 , representing human breathing during light workloads (Mineral Resources, 1994;Johnson, Weiss, & Grove, 1992). The electrical low pressure impactor (ELPI, TSI Inc./Dekati Ltd., St. Paul, MN, USA) was used to determine the concentration and aerodynamic particle size distribution in real-time. This instrument utilizes the cascade impaction principle and-in addition-has a direct-reading capability. The aerosol particles are charged by the corona charger downstream of the ELPI inlet and subsequently detected by the electrometers inside the cascade impactor. The particles were collected by the ELPI inside and outside the respirator using identical sampling lines. For these measurements, a 10-mCi Kr 85 charge equilibrator (3M Company, St. Paul, MN, USA) was installed upstream of the ELPI inlet to neutralize the particles to Boltzmann charge equilibrium. This allowed us to avoid the in uence of high electric charges, imparted by the particles as a result of their interaction with air ions, on the ELPI performance. The time resolution of the instrument was adjusted to 10 s. The data were recorded in 12 ELPI channels (each channel = impaction stage), from 0.04 to 8:4 m. The latter sizes represent the midpoint diameters of the ÿrst and the 12th impaction stages (the midpoint = the geometric mean of the stage's boundaries). The natural aerosol concentration in the indoor test chamber was not su cient, particularly for the measurement inside the mask, because the ÿlter removed considerable number of ambient airborne particles. To increase the initial background aerosol concentration, we used a smoke generator. The smoke particles covered primarily the submicrometer aerodynamic size range (Cheng, Bechtold, Yu, & Hung, 1995) with a sharp decrease in the particle number at d a ¿ 1:5-2 m. Thus, the data recorded in the ÿrst 8 measurement channels of the ELPI (d a = 0:04-1:3 m) were used for the analysis. The measured aerosol concentrations inside (C IN ) and outside (C OUT ) the mask were incorporated into the equation for the respirator penetration e ciency, E p : which was determined as a function of the particle aerodynamic diameter. E P is actually the inversed protection factor that is frequently used as a respirator performance index. Initially, the background tests were conducted by measuring the penetration e ciency of the mask with no ion emission. Then, a unipolar ion emitter was turned on at a distance of 20 cm from the mask, and E p (d a ) was determined in 3-min time intervals during 12 min. The continuous air ion emission in the chamber decreased C OUT as the particles, charged by ions to the same polarity, repelled and subsequently migrated toward the chamber's walls and deposited on these walls (Grinshpun et al., 2004). The change in the C OUT -value that occurred during each 3-min time interval due to ionic air puriÿcation in the chamber was taken into account through the linear interpolation of C OUT (t). In this study, we used a negative ion emitter (VI-2500, Wein, Inc., Los Angeles, CA, USA) producing an air ion concentration of N i ∼ 1:3 × 10 6 elementary charges per cm 3 as determined at a distance of 1 m from the source. The ion concentration was measured by the Air Ion Counter (AlphaLab Inc., Salt Lake city, UT, USA) that operates within the range of 10 -2×10 6 ions per cm 3 . Two types of ÿltering-facepiece respiratory masks commercially available from a major manufacturer were tested in this study. One was the NIOSH (US National Institute for Occupational Safety and Health) certiÿed N95 respirator and the other one was a conventional disposable surgical mask. The N95 respirator consists of inner and outer cover webs made of rayon. Its ÿlter made of polyester and polypropylene with the electrostatically charged microÿbers providing relatively high ÿltering e ciency. In the surgical mask, the polypropylene ÿlter is sandwiched between inner and outer webs made of rayon. The ÿlter of a surgical mask has lower ÿltration e ciency as compared to the one of an N95 respirator. Thus, the e ect of ion emission on the respirator ÿlter e ciency was tested for the masks having two distinctly di erent original performance characteristics. The average values and the standard deviations of the penetration e ciency were calculated for each set of conditions as a result of at least three replicates. The data were statistically analyzed using the Microsoft Excel software package (Microsoft Co., Redmond, WA, USA). Fig. 2 shows the normalized initial aerosol concentration measured outside the respirator mask. Each data point represents an average of six replicates. It is seen that the aerosol particles were primarily within a range of d a ≈ 0:04-0:5 m (the concentration N= log d a was between ¿ 10 4 and ¿ 10 5 cm −3 ), while fewer micron-size particles were detected ( N= log d a ∼ 10 3 cm −3 ). For each measured particle size, the initial aerosol concentration was reproducible with the variability (the coe cient of variation) not exceeding about 50% for six replicates. Fig. 3 presents the data obtained with two types of face-sealed respirators: N95 respirator and a surgical mask. When no air ion emission was introduced, the E p -value, averaged over the test range of d a , was about 1.8% for the N95 respirator. The particle size did not considerably a ect the penetration through the N95 respirator ÿlter, although some decrease of E p with increasing d a was observed for d a ≈ 0:04-0:5 m. Once the negative ion emission began, the penetration decreased to 0.27% during the ÿrst 3 min. At t = 12 min, it further decreased to about 0.11%, enhancing the N95 respirator ÿlter performance approximately by a factor of 17 (Fig. 3a). For the surgical mask, the initial penetration e ciency ranged from 18.7% (d a = 0:04 m) to 11.1% (d a = 1:3 m). Resulting from the emission of negative ions, the average penetration e ciency through the surgical mask dropped from 15.4% (t = 0) to 0.19% (t = 12 min), demonstrating a 80-fold enhancement (Fig. 3b). The data in Fig. 3 show that the most pronounced e ect occurred within the ÿrst 3-min interval. Baseline test It was surprising to observe that the initial penetration e ciency (t=0) of ultraÿne particles through both masks slightly increased with decreasing particle size. In contrast, the available ÿltration models predict that the peak penetration is reached at d a between 0.1 and 0:3 m, and the particles below 0:1 m should be collected more e ciently as their size decreases (di usion regime) (Halvorsen, 1998;Hinds, 1999;Lee & Mukund, 2001). The following considerations explain the results of our baseline test for the ultraÿne particles. These models (and the laboratory-generated experimental data that support them) characterize the particle penetration through a homogeneous perfectly sealed ÿbrous ÿlter material but not through a respirator mask. The design of a ÿltering-facepiece respirator does not assure a perfect peripheral connection of the assembly, so micro-leaks may be present between the core ÿlter material and the elastic peripheral support. These leaks can contribute to the penetration of the ultraÿne particles. In addition, although the mask was "glued" on the manikin, some very small, micrometer-or submicrometer-size leaks may still remain. Most of soap-bubble-based air leak detection methods are capable to identify micro-leaks greater than 1 m. If d a is much lower than the characteristic size of the micro-leaks (1 m), the particles may penetrate through these remaining submicrometer micro-leaks, thus a ecting the overall aerosol penetration e ciency through the ÿltering mask. One more possible factor is associated with the spatial variations in ÿber diameter, orientation, packing density, as well as initial ÿber electrostatic charge level (for those masks utilizing electret ÿlter media). These variations have been shown to signiÿcantly a ect the respirator performance increasing the penetration e ciency of particles of ∼ 0:1 m (Huang et al., 1998). Enhancement due to the unipolar air ion emission The enhancement of the respirator performance, observed almost immediately after the ion emitter started operating, can be attributed to the electrostatic e ect. The emitted negative ions as well as the particles charged by these ions in the air, impose signiÿcant negative charge on the respirator ÿlter. This forms the "electrostatic shield" against the particles moving toward the mask. The repelling forces decrease the number of particles that can approach the ÿlter. The above-described e ect works outside of the respirator, as opposite to the aerosol ÿltration by di usion, impaction, interception, and electrostatic deposition (Lee & Mukund, 2001) that takes place inside the ÿlter. Therefore, the ion-induced decrease in the particle penetration e ciency does not cause the pressure drop increase through the ÿlter providing the enhanced performance with the same comfort level. To quantitatively characterize the e ect, we calculated the velocity of particle migration induced by the electrostatic interaction in the vicinity of the ÿlter. It was then compared to the velocity of the air ow through the ÿlter caused by inhalation. In this calculation, we used information about the airborne particle electric charges and the air ion density, obtained by the ELPI and the Air Ion Counter, respectively. The particle size of d a = 0:1 m was chosen, as it represents the dominant size range used in our experiments. Furthermore, this value is at the boarder line between the ÿne and ultraÿne particle size ranges. In addition, many viral particles have an aerodynamic diameter of ∼ 0:1 m. Two main assumptions were made. First, the aerosol particles and air ions that interacted with the respirator were assumed to give all their electric charges to the respirator ÿlter. Second, the charged ÿlter was assumed to act as a point-charge located at the center of the respirator's surface. At the ion emission level produced in our experiment, a 0:1 m particle carries, on average, 10 elementary charges (Lee, Yermakov, & Grinshpun, 2004b). The calculation showed that the total electric charge acquired by the respirator ÿlter, as a result of a 3-min continued emission from the VI-2500 ion source, is about 1:5 × 10 13 elementary charges, if the breathing ow rate is 30 l min −1 . The ion concentration at the center of the respirator was determined to be approximately 1:6 × 10 8 cm −3 (this point located 20 cm from the ion emission source). Thus, the particle migration velocity was found to exceed the air ow velocity, created by the inhalation in the breathing zone, approximately by a factor of 75. This suggests that the e ect of the repelling force between the unipolarly charged particles and ÿlter surface is much stronger than the aerodynamic force. Therefore, although our assumptions may not be su ciently conservative, the above assessment demonstrates that the enhancement of the respirator performance by the unipolar ion emission is governed by the electrostatic "shield" mechanism. The drastic decrease of the particle penetration through the respirator ÿlter due to continuous unipolar ion emission may be critical in providing additional respiratory protection by existing masks against viral and bacterial particles. For example, an individual exposed to the in uenza virus concentration of 10; 000 m −3 inhales approximately 4500×0:18=810 viruses during 15 min when breathing through a conventional surgical mask at 30 l min −1 in the absence of ion emission (E p ≈ 18% for 0:1 m particles). The continuous emission of negative air ions (N i ∼ 10 6 e − cm −3 ) in a 25-m 3 room would reduce the indoor viral concentration by a factor of 9 during that 15-min interval (Lee et al., 2004a, b). In addition, it would enhance the surgical mask protection reducing E p from 18% to at least 0.19%. Thus, only about (4500=9) × 0:0019 = 0:95 ≈ 1 virus would be inhaled in 15 min. Given that the infectious dose of in uenza A2 is 790 viruses (Lawrence Berkeley National Laboratory), the ion emission e ect would make an important di erence with respect to the health risk. While this study is limited to the negative ion emission, we anticipate that the respirator performance enhancement e ect can be achieved also by generating positive air ions, as long as the ion concentration in the vicinity of the mask (breathing zone) is su ciently high. Future studies will address the e ects of polarity and the ion emission rate on the particle penetration e ciency through respirator ÿlters. Generally, the mask protection factor depends not only on its ÿlter penetration e ciency but also on its face ÿt (in practice, the facepiece mask is not sealed to the human face allowing the particles to penetrate through the leak). This pathway may become especially apparent when the ÿlter material is highly e cient. Therefore, future tests involving human subjects, di erent ÿt factors, and other experimental conditions ( ow rates and di erent masks) are needed to better characterize the enhancement e ect, discovered in this study, and link it to the exposure.	2018-12-27T08:07:46.949Z	2004-07-07T00:00:00.000	{ "year": 2004, "sha1": "ed1d74d5be1684827d20720b6881898f9c534f51", "oa_license": null, "oa_url": "https://doi.org/10.1016/j.jaerosci.2004.05.006", "oa_status": "BRONZE", "pdf_src": "PubMedCentral", "pdf_hash": "50d6f64f80fc7d304782b02c7669cbf4a6ae9d00", "s2fieldsofstudy": [], "extfieldsofstudy": [ "Medicine", "Materials Science" ] }
268864457	pes2o/s2orc	v3-fos-license	An overview of the clinical profile of infertile patients who underwent hysterosalpingography Background : Infertility is a prevalent medical issue that can arise from challenges in male or female or both partners. Pivotal contributing factors to female infertility include dysfunction, tubal issues, and peritoneal factors. Hysterosalpingography (HSG) holds significant importance in assessing abnormalities associated with the uterus and fallopian tubes. Aims and Objectives: This study aimed to assess the clinical profile and hysterosalpingographic findings of women with infertility who visited the department of radiodiagnosis for HSG. Materials and Methods : This is a cross-sectional study, where consecutive convenience-based sampling was used, with ethical clearance in place. Sixty-three infertile women of reproductive age were recruited for the study and were referred for HSG by the Department of Obstetrics and Gynecology. Results: The majority of the participants had primary infertility (71.88%) and was in the 31–35-year age group with the majority having 3–6 years of infertility and endometriosis. On HSG uterine pathology was found in 9.37% of participants and unilateral and bilateral tubal blocks were seen in 18.75% and 14.06% of cases approximately. Conclusion: HSG is a minimally invasive procedure, secure, and plays a crucial role in diagnosing uterine and tubal factors contributing to infertility. INTRODUCTION Infertility is a disease of the male or female reproductive system defined by the failure to achieve pregnancy within 12 months or more of regular unprotected sexual intercourse. 1 Infertility may be primary or secondary.Primary infertility has never conceived and secondary infertility indicates previous pregnancy irrespective of the outcome (abortion or live birth) but failure to conceive subsequently. 2Infertility is a global problem with frequency ranging from 5% to 15% in different communities.The United States sees a prevalence of 7.4-15.5% depending on the methodological approach used. 3India has a frequency of infertility of more than 10% approximately 20 million population as infertile as per data reproduced by (INITO -A Bangalore-based medical technology corporation).The causes of infertility with relative prevalence are as follows: Male factor (17-28%), both male and female factors (8-39%), female factor (33-40%), and unexplained (8-28%).The common male factors are defects of spermatogenesis (50%) and efferent duct obstruction (30%), epididymal obstruction (congenital or infective), ductal dysfunction or retrograde ejaculation, sperm motility disorder, sexual dysfunction, Erectile dysfunction, ejaculatory failure, drug-related, genitourinary surgery, 3 and unexplained. 4Female factors are responsible for approximately 40% of cases of infertility. 5The common causes of female factor infertility are ovarian causes, tubal disease, peritoneal causes, uterine, cervical factors, and hormonal causes. 6In the present era, a laparoscopy is an important tool in the infertility workup, which serves as both a diagnostic and therapeutic modality.It is reserved for selected cases to recognize endometriosis, pelvic adhesions, hydrosalpinx, and other peritoneal factors.As part of the infertility evaluation, adhering to guidelines the hysterosalpingography (HSG) becomes inevitable.Dyes are more appropriate in women with a history of pelvic inflammatory disease, previous ectopic pregnancy, or endometriosis.This allows assessment of both tubal and other pelvic pathology. 7Limited research on this subject has been done in our tertiary care hospital so far.Hence, it becomes imperative to conduct such a study in our hospital which caters to the most population of the rural strata. Aims and Objectives The aim and objective is to study the clinical profile of patients referred for HSG as a diagnostic tool in unexplained infertility. MATERIALS AND METHODS The present study was done in the Department of Radiology in collaboration with the Department of Obstetrics and Gynecology SKIMS MCH, Bemina, where 63 participants were recruited in the current study.After obtaining ethical clearance, well-informed consent was taken from the participants.Privacy and confidentiality were maintained.After a detailed history, clinical examination, and laparoscopic procedure, patients were referred for hysterosalpingography as part of the infertility evaluation protocol. A consecutive convenience-based sampling technique was used for the current study.After a thorough pre-invasive procedure, informed consent was taken from the patients.Patients were made to understand the procedure, advantages, complications, and need for HSG.After conducting a thorough history, initial assessment, and necessary investigations, they were asked to report post-menstrual in the proliferative phase for HSG as a part evaluation for infertility. Inclusion criteria • Patients having unexplained infertility in the reproductive age group • Patients having a history of primary as well as secondary infertility. Exclusion criteria • Patients having any contraindication to HSG procedures such as previous major surgeries, peritonitis, hernia large pelvic mass, or any medical comorbidity • Patients having contraindications to general anesthesia • Major male infertility. Procedure Patients were advised to take a mild pain reliever before the procedure and were informed about any allergies, particularly to iodine or contrast dye.The procedure was scheduled during the 1 st week after the menstrual period to minimize the risk of interference with a possible pregnancy.Patients were asked to lie on an examination table, usually on their back with their feet in stirrups, similar to a pelvic examination.The cervix is visualized by inserting a speculum inside the vagina.The cervix was cleaned with an antiseptic solution to reduce the risk of infection.A thin, flexible catheter was then gently inserted through the cervix and into the uterine cavity.A radiopaque contrast dye was injected through the catheter into the uterus.The dye helps to outline the uterine cavity and fallopian tubes, making them visible on X-ray images.The radiologist evaluated the images to identify any abnormalities, such as uterine abnormalities, blockages in the fallopian tubes, or other issues that may contribute to infertility.Once the procedure was complete, the catheter was removed.The patient was monitored for a short period to ensure that there were no immediate complications.Patients were provided with instructions regarding post-procedure care, including restrictions on activities or medications.Mild cramping or discomfort during or after the procedure was explained.Vaginal spotting or light bleeding after the HSG is common.Any signs of infection (fever, severe pain, etc.) were asked to be reported to the health-care provider. 7 Statistical analysis All the collected data were collected replicated in Microsoft Excel and analyzed using the IBM SPSS 24.0.Frequencies and percentages were calculated. RESULTS Table 1 indicates that in our study, primary infertility was found in 45 (71.42%) of the 64 patients, and secondary infertility was found in 18 patients 28.12%.Table 2 indicates that most patients in the primary infertility group were in the 31-35 years age group and understandably 34.3% from 26 to 30 years followed by 7.8% from the age group of 36-40 years.Table 3 indicates that the majority of patients 65.07% had 3-6 years of duration of illness, followed by 25.39% who had a duration of fewer than 2 years and 9.52% had >6 years of duration.Table 4 and Figures 1-3 indicates that the majority of patients having primary infertility had endometriosis (51.51%), pelvic adhesions 18.18%, tubal block 12.12%, myoma 9.09%, and hydrosalpinx 9.09%, whereas in secondary infertility, majority 50.00% had endometriosis, 37.5% had pelvic adhesions, and the tubal block was seen in 12.5%.The overall majority of patients had endometriosis 51.21%. Table 5 and above HSG images indicate that a unicornuate uterus was found in 3.12%, an irregular shape of uterine cavity was found in 9.37%, unilateral tubal block was seen in 18.75%, and bilateral in 12.50%. DISCUSSION The study was done to ascertain the clinical profile of infertility patients undergoing HSG.The study was done from June to October 2023.Our study was a crosssectional hospital-based study.A total of 63 infertile patients were evaluated.The majority of patients were from the age group of 26 to 35 years in our study which was done by study. 8Our study documented a little higher percentage of primary infertility than the study done by Khan et al. 9 The likely reason could be influenced by different characteristics of the study population and the definition of abnormalities.The majority of patients had a duration of infertility between 2 and 5 years which where the mean duration of infertility was >5 years. 10ur study documented that the majority of patients had endometriosis in laparoscopic findings which was in concordance with the study. 11,12In our study, the results showed that a significant number of participants had a unilateral and bilateral tubal block, similar findings have been explained in studies by Bello, Tvarijonaviciene, Tsuji et al. 12,13,14 Different procedures are adopted to assess and evaluate the infertile population of reproductive age but still, HSG remains a good choice to investigate the uterine and tubal cause of infertility. What this study adds? • This study may provide diagnostic insight and treatment planning • Improvement initiatives in health-care settings, encourage the refinement of diagnostic protocols and the enhancement of patient care for infertile individuals • May foster collaboration between different medical specialties involved in infertility management, promoting a comprehensive and multidisciplinary approach to patient care. Limitations of the study 1.The smaller sample size limits it to make it generalized 2. The sample was taken from a single tertiary care hospital 3. The study has recruited from a sample from similar ethnic or cultural populations 4. Need for a follow-up study to understand the longterm outcomes of infertile patients after HSG. CONCLUSION Our study provides an important overview of the clinical profile of infertile patients undergoing hysterosalpingography (HSG).Our findings reveal that a major proportion of these patients present with tubal blockages unilateral or bilateral, uterine abnormalities, and endometriosis.Our results highlight the importance of HSG in diagnosing and understanding the underlying causes of infertility.Extensive research is needed to explore the impact of HSG findings on treatment outcomes and to optimize management strategies for the infertile population.	2024-04-03T15:58:01.053Z	2024-04-01T00:00:00.000	{ "year": 2024, "sha1": "69c7fcf0912b5f4be46f8e4bfec74b8aee800d1a", "oa_license": "CCBYNC", "oa_url": "https://www.nepjol.info/index.php/AJMS/article/download/62106/48293", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "3761a9bbbef6131e3326176354d3a4d0e967bad6", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [] }
213226869	pes2o/s2orc	v3-fos-license	Independent and sensitive gait parameters for objective evaluation in knee and hip osteoarthritis using wearable sensors Background Although it is well-established that osteoarthritis (OA) impairs daily-life gait, objective gait assessments are not part of routine clinical evaluation. Wearable inertial sensors provide an easily accessible and fast way to routinely evaluate gait quality in clinical settings. However, during these assessments, more complex and meaningful aspects of daily-life gait, including turning, dual-task performance, and upper body motion, are often overlooked. The aim of this study was therefore to investigate turning, dual-task performance, and upper body motion in individuals with knee or hip OA in addition to more commonly assessed spatiotemporal gait parameters using wearable sensors. Methods Gait was compared between individuals with unilateral knee (n = 25) or hip OA (n = 26) scheduled for joint replacement, and healthy controls (n = 27). For 2 min, participants walked back and forth along a 6-m trajectory making 180° turns, with and without a secondary cognitive task. Gait parameters were collected using 4 inertial measurement units on the feet and trunk. To test if dual-task gait, turning, and upper body motion had added value above spatiotemporal parameters, a factor analysis was conducted. Effect sizes were computed as standardized mean difference between OA groups and healthy controls to identify parameters from these gait domains that were sensitive to knee or hip OA. Results Four independent domains of gait were obtained: speed-spatial, speed-temporal, dual-task cost, and upper body motion. Turning parameters constituted a gait domain together with cadence. From the domains that were obtained, stride length (speed-spatial) and cadence (speed-temporal) had the strongest effect sizes for both knee and hip OA. Upper body motion (lumbar sagittal range of motion), showed a strong effect size when comparing hip OA with healthy controls. Parameters reflecting dual-task cost were not sensitive to knee or hip OA. Conclusions Besides more commonly reported spatiotemporal parameters, only upper body motion provided non-redundant and sensitive parameters representing gait adaptations in individuals with hip OA. Turning parameters were sensitive to knee and hip OA, but were not independent from speed-related gait parameters. Dual-task parameters had limited additional value for evaluating gait in knee and hip OA, although dual-task cost constituted a separate gait domain. Future steps should include testing responsiveness of these gait domains to interventions aiming to improve mobility. Supplementary Information The online version contains supplementary material available at 10.1186/s12891-021-04074-2. (Continued from previous page) Conclusions: Besides more commonly reported spatiotemporal parameters, only upper body motion provided nonredundant and sensitive parameters representing gait adaptations in individuals with hip OA. Turning parameters were sensitive to knee and hip OA, but were not independent from speed-related gait parameters. Dual-task parameters had limited additional value for evaluating gait in knee and hip OA, although dual-task cost constituted a separate gait domain. Future steps should include testing responsiveness of these gait domains to interventions aiming to improve mobility. Keywords: Knee osteoarthritis, Hip osteoarthritis, Inertial measurement units, Gait analysis, Dual-task Background It is well-recognized that osteoarthritis (OA) of the knee or hip impairs gait [1][2][3][4]. Indeed, individuals with knee or hip OA walk less during daily life and their quality of gait is compromised [5]. Yet, objective gait assessments are not part of routine clinical evaluation, and gait difficulties in OA are insufficiently captured by patientreported outcome measures [6][7][8]. In part, this may be due to limited time available during clinical visits, considering that gait analysis is traditionally conducted in a gait laboratory, making it time consuming and not easily accessible. Recent advances in inertial sensor technology have opened up new avenues to quickly and objectively assess gait quality in a clinical setting. Small inertial measurement units (IMUs) can be used to quickly and accurately obtain gait parameters without being restricted to a fixed (laboratory) environment [9,10]. Moreover, compared to gait analysis in a lab, substantially more strides can be collected in a shorter period of time. On the downside, an important issue of gait assessment with IMUs is that it typically results in a large number of outcome parameters, with numerous correlated parameters. For example, many gait parameters share covariance with gait speed [11][12][13][14][15]. Hence, for clinical implementation, it is important to identify gait parameters from independent gait domains that best describe the gait adaptations in individuals with knee and hip OA compared to healthy controls. So far, ambulatory gait assessments in individuals with knee and hip OA have mostly been limited to simple, straight-ahead walking paradigms [16]. Parameters reflecting more complex and relevant aspects of gait, including dual-task gait, turning, and compensatory trunk motion are less frequently reported in studies using IMUs. Turning and dual-task performance have been shown to be important aspects of daily life ambulation in elderly populations and can easily be assessed using wearable sensors [17][18][19][20]. Turning is a common cause of falling in community dwelling elderly, and may be more sensitive to sensorimotor impairments than straight-ahead gait [19,21]. Dual-task performance, on the other hand, reflects the amount of attentional resources allocated to gait [22]. In order to compensate for gait difficulties caused by OA, a strategy could be to allocate more attention to gait. The extent to which a secondary cognitive task affects gait performance (i.e. dual-task cost (DTC)) may therefore be larger in individuals with OA. A recent scoping review indicated that DTC was not different between individuals with knee OA and healthy controls during quiet standing and forward induced falls [23]. However, DTC during gait has not yet been compared between those groups. A third gap in literature regarding wearable sensors and OA is the lack of attention for upper body movement. Upper body motion is important for maintaining stability, but may also be indicative of compensatory gait changes that reflect OA-related pain or disability [24][25][26]. The aim of this study was therefore to investigate turning, dual-task performance, and upper body motion in addition to spatiotemporal gait parameters in individuals with knee or hip OA, taking shared covariance between gait parameters into account. More specifically, we aim to test if 1) turning, dual-task gait, and upper body motion constitute independent domains of gait in our sample, and 2) gait parameters in these gait domains can discriminate individuals with knee or hip OA from healthy controls. Together, these findings may contribute to a better understanding of the multidimensional aspects of gait, and how this is affected in knee and hip OA. Participants In this cross-sectional, comparative study 78 participants were included. The total study population comprised three groups: individuals with unilateral knee OA (n = 25), unilateral hip OA (n = 26), and healthy controls (n = 27). Samples were derived from a longitudinal study investigating gait before and after total knee and hip arthroplasty that was powered for the comparison of spatiotemporal gait characteristics between individuals 1 year after total knee or hip arthroplasty and healthy controls. Individuals with OA were recruited at the Sint Maartenskliniek and were included if they had both radiological and symptomatic OA and were listed for joint replacement surgery. Participants had to be able to walk for more than 2 min without the use of any assistive device. Exclusion criteria were: 1) expectancy of joint replacement within a year, or symptomatic OA, in another weight-bearing joint than the joint scheduled for surgery, 2) BMI > 40 kg/m 2 , and 3) any other musculoskeletal or neurological impairment interfering with gait or balance. Healthy controls were recruited from the community and did not have a clinical diagnosis of knee or hip OA, nor did they have any pain in the lower extremities. Healthy controls were recruited in the same age range as individuals with OA. Exclusion criteria for healthy controls were the same as for individuals with knee and hip OA. Informed consent was obtained from all participants prior to testing. Ethical approval was obtained from the CMO Arnhem/Nijmegen (2018-4452). All study methods were carried out in accordance with the Declaration of Helsinki. Demographic and clinical assessment Evidence for radiological OA was provided by the Kellgren and Lawrence (KL) score as assessed by experienced orthopedic surgeons [27]. Anthropometric characteristics were obtained during the pre-operative screening visit and were summarized as mass, length, and BMI. For individuals with knee and hip OA, self-reported functioning was assessed using the Knee Injury and Osteoarthritis Outcomes Score (KOOS) or Hip Disability Osteoarthritis Outcome Score (HOOS) [28,29]. All items were scored on a zero to four Likert scale. For the five subscales, total scores were transformed to a 0-100 scale, with 100 representing best function. Gait assessment Gait parameters were collected on the same day as the pre-operative screening visit, which took place approximately 1 to 2 months prior to surgery. Four IMUs (Opal V2, APDM Inc., Portland, OR) were used to obtain segment accelerations and angular velocities (sample frequency = 128 Hz). Sensors were attached via elastic straps to the dorsum of both feet, the waist (sacrolumbar level), and the sternum ( Fig. 1) according to the standardized sensor placement of MobilityLab. Participants walked wearing flat shoes at a self-selected comfortable speed. For a duration of 2 min, participants walked back and forth along a 6-m trajectory making 180°turns (Fig. 1). Two 2-min trials were collected, with and without a secondary cognitive task. The cognitive task consisted of an alternating alphabet task, citing every other letter of the alphabet. Single-task walking was always performed before the dual-task condition. Responses to the cognitive task were recorded by the assessor. Accuracy on the cognitive task was summarized as correct responses (percentage of total responses). DTC was computed as the percentual change of dual-task performance relative to the single-task for the following parameters: gait speed, cadence, stride length, stride time variability, and turn duration. Data analysis Gait parameters were extracted from the raw IMU signals using the commercially available and validated Mobility Lab v2.0 software package [30]. Mobility Lab uses a state space model with causal Kalman filter along with zero velocity updates for optimal orientation estimation. Range of motion metrics were described for both the lumbar and trunk sensors using the gyroscope signals. As such, these measures are representative of the rotation of the sensors, which is caused by the movement of the underlying segments. For parameters where side was relevant (i.e. foot elevation at midswing, lateral step variability, circumduction, foot strike angle, toe off angle, and stance duration), we analyzed the affected leg in Fig. 1 Overview of the experimental set-up. Four IMUs were attached to the dorsum of both feet, lumbar level (L4/L5) of the waist, and the sternum. For 2 min, subjects walked back and forth over the 6 m trajectory, making 180 degree turns individuals with knee or hip OA, whereas for healthy controls the average value from the left and right leg was taken. Gait parameters were initially selected based on reliability, theoretical considerations, and completeness (< 20% missing values). Based on the reliability criterium, we excluded stance and swing duration as percentage of gait cycle [31]. With regard to theoretical considerations, the following decisions were made: 1) in case gait parameters reflected the same outcome (e.g. gait cycle duration and cadence) only one parameter was kept for further analysis, 2) asymmetry parameters were restricted to meaningful parameters (i.e. stride length cannot be asymmetric when walking over a straight path) [32]. DTC of gait parameters that are ratios (i.e. asymmetry values) were not included in order to prevent inflated values, except for stride time variability, due to the substantial number of other studies evaluating this parameter in the context of fall risk [33]. This resulted in twenty-five gait parameters entered into factor analysis to identify correlated outcomes. Exploratory factor analysis was used to identify independent gait domains explaining the variance in gait performance. Adequacy of the dataset for factor analysis was tested using Barlett's test of sphericity and the Kaiser-Meyer-Olkin (KMO) test. In case individual KMO values were lower than 0.5, variables were removed from the analysis [34]. The number of factors to be retained for further analysis was determined using the Kaiser criterium (eigenvalue > 1.0) [35]. Subsequently, factor analysis with varimax rotation was performed to obtain orthogonal factor scores. Within a factor, gait parameters were considered relevant when they met a minimum factor loading of 0.5. For each relevant gait parameter in the obtained factor, effect sizes were computed as standardized mean differences (SMD) for the comparison between the OA groups and healthy controls (knee OA vs healthy controls and hip OA vs healthy controls). The gait parameter with the highest factor loading in combination with an effect size larger than 0.5 was considered nonredundant and sensitive to either knee or hip OA. For these gait parameters, individual datapoints and means with 95% confidence intervals (CI) were constructed using estimation graphs to assess between-group differences [36]. For demographic and clinical parameters, main group effects (3 levels: knee OA, hip OA, healthy controls) were tested using a one-way ANOVA or non-parametric equivalent when assumptions for parametric testing were not met. In case of a significant main effect, a posthoc comparison was conducted using independent samples Student's t-test or the non-parametric equivalent. Data was considered statistically significant at an alpha level of 0.05, which was adjusted for multiple comparisons (n = 9) for the gait parameters. This resulted in a Bonferroni adjusted alpha level of 0.0056. Data analysis was performed using STATA and customwritten Python scripts incorporating the DABEST library [37]. Exploratory factor analysis Twenty-five gait parameters were entered into the factor analysis (Fig. 2). Based on individual KMO values, the following variables were removed from further analysis: DTC of stride length, trunk transverse range of motion (RoM), lateral step variability, toe-out angle, and foot elevation at midswing. Factor analysis of the remaining twenty parameters yielded four orthogonal factors accounting for 87.8% of the total variance in gait performance ( Table 2). The factors were described as speedspatial, speed-temporal, dual-task cost, and upper body motion. Gait speed had a cross-loading on the factors speed-spatial (0.759) and speed-temporal (0.579). Turning parameters loaded on the factor speed-temporal. In the upper body motion domain, factor loadings of the parameters were relatively low, ranging between 0.53 and 0.61. Selection of gait parameters based on effect size SMDs for the comparison between OA groups and healthy participants are visualized for all gait parameters in Fig. 3. Based on the criterium for effect size, the following gait parameters were selected to represent the corresponding factors: stride length (speed-spatial), cadence (speed-temporal), and lumbar sagittal RoM (upper body motion). Although the factor DTC explained 20.7% of the total variance in gait performance, none of the gait parameters within this factor showed an effect size larger than 0.5 (Fig. 3). Gait speed showed the largest effect size for both the comparison between knee OA and controls (SMD = 1.59) and hip OA and controls (SMD = 1.70). However, due to cross-loadings on factors speedspatial and speed-temporal, gait speed was not prioritized over stride length and cadence. In addition, many of the gait parameters from the factor speed-spatial and speed-temporal showed large effect sizes (SMD > 0.8) for both group comparisons. Between group comparisons of non-redundant gait parameters Between-group differences of the selected gait parameters were visualized using estimation plots (Fig. 4). Both individuals with knee and hip OA walked with a lower cadence and with shorter steps. More specifically, compared to healthy controls stride length was 0.17 m (95% CI: 0.09-0.26, p < 0.001) lower in individuals with knee OA and 0.20 m (95% CI: 0.12-0.28, p < 0.001) lower in hip OA. In addition, cadence was 10.8 steps/min (95% CI: 6.3-15.4, p < 0.001) lower in individuals with knee OA and 9.8 steps/min (95% CI: 5.2-14.4, p < 0.001) lower in individuals with hip OA. Lumbar RoM in the sagittal plane was 2.7 degrees (95% CI: 1.7-4.4, p < 0.001) higher for individuals with hip OA compared to controls, whereas no differences were found between knee OA individuals and healthy controls (mean difference = 0.5 degrees, 95% CI: − 0.33-1.59, p = 0.260). Discussion The aim of the present study was to investigate turning, dual-task performance, and upper body motion in addition to spatiotemporal gait parameters in individuals with knee or hip OA. To avoid redundancy of gait parameters, we conducted a factor analysis. Four independent gait domains were identified: speed-spatial, speedtemporal, dual-task cost, and upper body motion. Turning did not constitute its own domain but was related to speed-temporal. Three domains held parameters sensitive to knee or hip OA: speed-spatial (stride length), speed-temporal (cadence), and upper body motion (lumbar sagittal RoM). Dual-task cost was not sensitive to knee or hip OA. Factor analysis effectively reduced the dimensionality of our dataset from twenty-five gait parameters to four independent domains of gait, including domains related to dual-task gait and compensatory trunk motion. Turning, however, was part of a factor together with cadence. The factors explaining most of the variance in our sample, i.e. speed-spatial and speed-temporal, were both dependent on gait speed ( Table 2). In the literature, these factors reflecting the spatial and temporal aspects of gait speed are consistently reported [38][39][40][41][42]. Other factors related to gait are variability [38,39,41,42], asymmetry [39,41,42], postural control [39], and trunk motion [40]. Dual-task cost has not previously been evaluated in a factor analysis approach, but may contain unique information about gait that is informative of disease-specific compensations related to the reallocation of attentional resources. Importantly, dualtask cost and upper body motion are interesting domains as they were independent of gait speed, evidenced by the absence of a cross-loading of gait speed on these domains in our study. Dual-task cost and upper body motion may therefore provide promising gait parameters for clinical evaluation of gait, in addition to the more commonly used speed-related measures. In our analysis steps, turning parameters were excluded in favor of the stronger factor loading that was obtained for cadence. However, effect sizes for turning were large when comparing both knee and hip OA with healthy participants (SMD > 0.9, Fig. 3). In addition, factor loadings were not substantially lower compared to cadence. Taken together, we are unsure whether this factor represents a combination of gait and turning, or better reflects turning itself. Future research should therefore indicate as to what extent turning parameters Fig. 2 Flowchart describing the selection process of gait parameters. Note: foot elevation at midswing = height of the foot sensor at mid-swing, lateral step variability = spatial deviation in the lateral direction of each foot compared to previous steps, circumduction = amount that the foot travels perpendicular to forward movement during the swing phase are driven by cadence or gait speed, and how meaningful the unexplained variance is for evaluation of physical functioning in individuals with knee and hip OA. To facilitate assessment of the between-group differences, we opted to select single gait parameters from the independent factor, to represent the respective factor. From the factors that we obtained, only dual-tasking parameters did not discriminate between knee or hip OA and healthy controls (SMD < 0.5). This indicates that, compared to healthy controls, individuals with OA did not need more attentional resources for the motor task. Thus, although gait was affected in OA, this was not compensated by more attentional resources. Many of the gait parameters that showed large between-group effect sizes (Fig. 3) were grouped either under the speed-spatial or under the speed-temporal domain. This suggests that the two main components determining gait speed, stride length and cadence, are inherently linked with various gait adaptations prominent in individuals with knee and hip OA. As such, gait speed may also be considered as the final common pathway for various gait adaptations, and could be used as a Red colors indicate OA < healthy controls, green colors represent OA > healthy controls. Please note that gait speed had a cross loading and was also part of the speed-temporal domain. Note: CV = coefficient of variation, DTC = dual-task cost, RoM = range of motion very general, but highly sensitive marker for functional status in individuals with OA. Next to this, our findings further stress the need to take gait speed differences into account when evaluating gait in individuals with OA. More specifically, for parameters that are correlated with gait speed, it may be more appropriate to assess them at a standardized, matched speed, as it may be difficult to separate effects of gait speed from the effects of OA itself [43]. Finally, these findings underline the importance of data reduction techniques when investigating gait using IMUs or motion capture systems, as statistical testing of all gait parameters would increase the probability of finding false positives. That speed-related gait parameters have good discriminatory capacity in OA has been reported before. Two systematic reviews reported lower gait speed and stride length in individuals with knee and hip OA compared to healthy participants [1,3]. In studies employing IMUs, similar changes in stride length and cadence were found [25,44]. In absolute numbers, slight differences with our values can be discerned. Reasons for this may include the relatively short walking distance (6 m) in this study that was necessary to reliably assess turning, versus the longer distances (~20 m) that are commonly used. Nevertheless, our findings corroborated previous findings about the discriminatory capacity of stride length and cadence. In addition to spatiotemporal differences, individuals with hip OA walked with distinct upper body motion, which was most evident in the sagittal plane at the lumbar level. However, upper body motion is difficult to capture by just one parameter, as is illustrated by the relatively low factor loadings lying close together in this domain (Table 1). Altered trunk motion may point toward the use of compensatory strategies to unload the arthritic joint [45]. More specifically, increased pelvic RoM in the sagittal plane may enable more effective anteflexion of the lower limbs and may thereby, to a certain extent, preserve stride length [46]. In addition, anterior pelvic tilt combined with lateral trunk lean can reduce the lever arm between the hip joint center and center of mass [25]. We observed more lumbar sagittal RoM and more RoM of the trunk in the coronal plane in individuals with hip OA compared to healthy controls, in line with previous reports [25,46]. Unfortunately, the exact reason for the use of these compensatory mechanisms remains speculative and may relate to pain, muscle weakness, or joint instability [47]. Future research should therefore investigate the importance of upper body motion in individuals with OA, to inform us about potential mechanisms underlying these gait adaptations. With regard to the use of wearable sensors in clinical practice, our study showed that quick and easy gait assessments with wearable sensors are useful for evaluating gait impairments in individuals with knee and hip OA. In comparison to optical motion capture systems, wearable sensors are more feasible for large-scale use and could be utilized to routinely assess physical functioning. From all gait parameters, gait speed was found to be a very general but highly sensitive marker for mobility limitations, combining both the effects on stride Fig. 4 Estimation plots of the mean group differences for stride length, cadence, and lumbar sagittal RoM. In the top panel, dots represent the individual datapoints and bars the mean (± SD). In the bottom panel, the distribution of the mean difference (± 95% CI) for the comparison with healthy controls is visualized. In cases where zero is not in the 95% CI of the mean difference, as indicated by the black bars in the lower panels, data was statistically different at p < 0.05 length and cadence. Besides the basic spatiotemporal measures, trunk motion and turning appeared to be relevant for individuals with knee and hip OA. We therefore recommend to use sensor configurations that allow to look beyond these basic spatiotemporal parameters. In the future, wearable sensors should also be utilized to their fullest potential to enable remote monitoring at home, which would allow to more accurately capture the habitual gait patterns. This study had several limitations that merit attention. First, we did not obtain factors representing gait asymmetry or variability, which may have been related to the low number of gait parameters related to those domains that were initially entered into factor analysis. We were therefore limited in our conclusions regarding the potential value of those measures for individuals with knee or hip OA. Second, five potentially valuable gait parameters were removed from further analysis due to sampling inadequacy (KMO value < 0.5). Larger sample sizes are therefore required to identify the potential value of these parameters. Related to this, we did not include demographic or clinical variables in the factor analysis, as this could have affected the accuracy of factor analysis due to the relatively small sample size. Finally, including individuals with isolated, unilateral knee or hip OA was important for our study purposes, although the majority of the OA population have complaints in more than one joint [48]. We expect that widening the inclusion criteria would have resulted in larger differences of OA groups compared to healthy controls, but in less specificity for each OA group. In addition, it is important to note that individuals in this study had end-stage OA and were scheduled for joint replacement. Our results may thus not be representative of gait in individuals with less severe OA. Conclusion In addition to commonly assessed spatiotemporal parameters, this study provided two other relevant domains of gait: dual-task cost and upper body motion. Although dual-task cost provided unique information about gait, our results did not suggest that individuals with knee or hip OA needed more attention for walking than healthy participants. Adaptations in upper body motion were more subtle than stride length and cadence, but may carry important information about compensatory strategies that are most distinctive for individuals with hip OA. Future steps should include evaluation of the responsiveness of these gait parameters to effects of interventions aiming to improve mobility, such as joint replacement surgery. Furthermore, longitudinal monitoring of individuals with knee and hip OA starting at earlier stages of the disease may inform us about the development of these gait adaptations and associated compensations over time.	2020-02-27T09:32:00.324Z	2020-02-21T00:00:00.000	{ "year": 2021, "sha1": "1b1e143bf525ec0865a0689c454cc8379c9832dc", "oa_license": "CCBY", "oa_url": "https://bmcmusculoskeletdisord.biomedcentral.com/track/pdf/10.1186/s12891-021-04074-2", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "79b94f3ea4e4e312d404918ae9cd1cba46f88e81", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine", "Biology" ] }
247962446	pes2o/s2orc	v3-fos-license	Introducing a Chemically Intuitive Core-Substituent Fingerprint Designed to Explore Structural Requirements for Effective Similarity Searching and Machine Learning Fingerprint (FP) representations of chemical structure continue to be one of the most widely used types of molecular descriptors in chemoinformatics and computational medicinal chemistry. One often distinguishes between two- and three-dimensional (2D and 3D) FPs depending on whether they are derived from molecular graphs or conformations, respectively. Primary application areas for FPs include similarity searching and compound classification via machine learning, especially for hit identification. For these applications, 2D FPs are particularly popular, given their robustness and for the most part comparable (or better) performance to 3D FPs. While a variety of FP prototypes has been designed and evaluated during earlier times of chemoinformatics research, new developments have been rare over the past decade. At least in part, this has been due to the situation that topological (atom environment) FPs derived from molecular graphs have evolved as a gold standard in the field. We were interested in exploring the question of whether the amount of structural information captured by state-of-the-art 2D FPs is indeed required for effective similarity searching and compound classification or whether accounting for fewer structural features might be sufficient. Therefore, pursuing a “structural minimalist” approach, we designed and implemented a new 2D FP based upon ring and substituent fragments obtained by systematically decomposing large numbers of compounds from medicinal chemistry. The resulting FP termed core-substituent FP (CSFP) captures much smaller numbers of structural features than state-of-the-art 2D FPs. However, CSFP achieves high performance in similarity searching and machine learning, demonstrating that less structural information is required for establishing molecular similarity relationships than is often believed. Given its high performance and chemical tangibility, CSFP is also relevant for practical applications in medicinal chemistry. Introduction Similarity searching using molecular fingerprints (FPs) has a long history in chemoinformatics, especially for computational hit identification [1][2][3][4][5][6][7]. In similarity searching, the chemical similarity between query and database compounds is quantified on the basis of FP overlap and used to infer property similarity (such as a similar biological activity) [3][4][5]. FPs include bit string representations of chemical structures and/or molecular properties, as well as feature sets [1][2][3]. Various FP designs have been introduced to encode three-dimensional (3D) molecular features such as pharmacophore patterns or two-dimensional (2D) features such as substructures [1,5,[8][9][10]. In addition, the bit string FP format has also been used to encode protein-ligand interactions [11,12]. By design, 2D FPs are simpler than 3D FPs but by no means less effective in detecting molecular similarity relationships and identifying new active compounds (2D FPs are typically less sensitive to feature noise than 3D FPs) [1][2][3][4][5]. Despite their conceptual simplicity, 2D FPs have often been surprisingly successful in identifying structurally diverse compounds with desired biological activities [2][3][4]. In general, 2D FPs include both keyed and hashed formats [2,3]. In keyed representations, each bit position corresponds to the presence or absence of a specific feature (or, albeit much less frequently used, a feature count). In hashed FPs, features are mapped to overlapping bit segments (hence producing specific bit patterns without 1:1 bit-to-feature correspondence). Accordingly, a bit position might be set on by different features (a phenomenon referred to as bit collision). Hashing algorithms can also be applied to transform molecule-specific feature sets into a bit string format of constant length. Furthermore, 2D FPs can be classified into two major and a few minor categories. The two major categories account for substructure-based FPs and others capturing topological patterns, as further discussed below. In addition, FPs have been introduced to encode 2D pharmacophore patterns [13,14], values of numerical 2D property descriptors [15], or combinations of binarized numerical descriptors and structural fragments [15,16]. For the major class of substructure-based FPs, molecular access system (MACCS) structural keys have been a pioneering prototype [17,18]. As the name implies, the design of these FPs is keyed and their predefined substructures include structural fragments and rule-based structural (SMARTS) patterns [18]. MACCS FP versions including 960 or 166 structural keys have been introduced, but the smaller 166-bit version has often met or even exceeded the performance of the larger one in similarity searching and has become a standard keyed FP design. Structural redundancy in such FPs often increases the noise in similarity calculations, especially for chemically complex molecules, and rational bit reduction strategies [16] or even random bit removal [19] have been applied to stabilize or further increase search performance. Other substructure-based FPs include variably sized versions of BCI FPs based upon a catalog of 1052 fragments [20] and the PubChem FP comprising 881 structural keys [21]. In addition, different types of atom pair-based FPs have been introduced [22][23][24]. However, none of these FP designs has replaced MACCS keys as a standard for substructure-based similarity searching. The other major class of 2D FPs comprises FPs capturing different types of topological patterns. One pioneering development has been the suite of daylight FPs that represent hashed designs with a length of up to 2048 bits and account for connectivity pathways through molecules [25]. The second class of topological FPs conceptually originates from the Morgan FP [26], another pioneering development, and includes various circular atom environment FPs [27][28][29]. Among these, extended connectivity FPs (ECFPs) capturing layered atom environments of varying bond diameters (e.g., four; corresponding to ECFP4) [29] have become a standard in the field, due to their typically highest search performance in 2D FP benchmarking [3,5,7]. ECFPs are directly based upon the Morgan algorithm and represent molecule-specific sets of layered atom environments. However, they are mostly used as constantly sized keyed or hashed bit strings, with ECFP4 being the most widely applied representative. In addition, ECFP variants with feature counts or feature groupings according to similar atom functions have been introduced [29] but are much less frequently used than ECFP4. Since many layered atom environments in compounds overlap, a characteristic of ECFPs is intrinsic topological feature redundancy, which (different from redundancy in some substructure FPs) does not notably affect search performance. Given its typically high performance, ECFP4 has become a gold standard for 2D similarity searching and is currently the most widely used FP. Over the past decade, with ECFPs becoming a mainstay in the field, conceptually new designs of FPs have been rare. In this study, we report a new 2D FP with lower feature numbers and bit density than commonly used FPs. This new FP was specifically designed to explore the question of how much structural information might be required for effective similarity searching. Exploring this question was inspired by earlier attempts to generate "mini-fingerprints" through reductionist approaches [16], as well as observations that, for given compound classes, even small sets of substructures from randomly generated populations might yield predictive FPs [30]. While dependent on specific classes of active compounds, these earlier observations partly inspired our current study, aiming to obtain further general insights into structural requirements for similarity searching. The new FP reported herein, termed core-substituent FP (CSFP), is keyed and particularly intuitive from a medicinal chemistry perspective. In addition to serving as a research tool for similarity searching and machine learning, the performance of CSFP compared with MACCS and ECFP4 observed in our study also indicates its potential for practical applications. Fingerprint Design Principles The general idea underlying the design of CSFP was the generation of an easily interpretable FP consisting of molecular fragments for representing as many compounds as possible with as little structural information per compound as possible. Hence, following a "structural minimalist" approach, as reflected by the number of structural features detected in a compound, a keyed substructure FP was envisioned comprising ring fragments from compound cores plus substituents, as illustrated in Figure 1, representing a new FP design strategy. For a given compound, we aimed to obtain fewer decomposed fragments than structural keys or atom environments, hence yielding a generally applicable FP with a smaller number of features than MACCS or ECFP4. Therefore, as a basis for CSFP design, a new fragment library was generated. classes, even small sets of substructures from randomly generated populations might yield predictive FPs [30]. While dependent on specific classes of active compounds, these earlier observations partly inspired our current study, aiming to obtain further general insights into structural requirements for similarity searching. The new FP reported herein, termed core-substituent FP (CSFP), is keyed and particularly intuitive from a medicinal chemistry perspective. In addition to serving as a research tool for similarity searching and machine learning, the performance of CSFP compared with MACCS and ECFP4 observed in our study also indicates its potential for practical applications. Fingerprint Design Principles The general idea underlying the design of CSFP was the generation of an easily interpretable FP consisting of molecular fragments for representing as many compounds as possible with as little structural information per compound as possible. Hence, following a "structural minimalist" approach, as reflected by the number of structural features detected in a compound, a keyed substructure FP was envisioned comprising ring fragments from compound cores plus substituents, as illustrated in Figure 1, representing a new FP design strategy. For a given compound, we aimed to obtain fewer decomposed fragments than structural keys or atom environments, hence yielding a generally applicable FP with a smaller number of features than MACCS or ECFP4. Therefore, as a basis for CSFP design, a new fragment library was generated. For our analysis, we distinguished between core and substituent fragments. For the generation of core fragments, analogue series (ASs) were algorithmically extracted isolated from compounds from medicinal chemistry sources, as specified in the Materials and Methods Section, and their core structures were sampled. Obtaining core structures from ASs ensured that the cores were represented by multiple compounds and were thus generalizable. All extracted cores contained ring structures, and these cores were then decomposed into fused and single rings, as described below. The resulting core fragments were complemented with an equally sized set of most frequently occurring substituent fragments selected from an R-group replacement database we have recently generated and made publicly available [31], resulting from a new methodology for systematically extracting R-groups from compounds [32]. Core Structure Fragmentation Accounting only for the core structure of a compound would produce a single feature, which is of course not suitable for FP design. Therefore, from core structures, all fused and single rings were extracted first by the removal of single bonds between ring systems. Subsequently, fused rings were decomposed into individual ring fragments. All single and fused rings extracted from each core were recorded. Figure 2a-d illustrate the systematic extraction of fused and single rings from cores of exemplary compounds belonging to different ASs (in these cases, removal of all single-bonded substituents from rings yields the core). Importantly, for ring fragments extracted from fused rings, atomic hybridization states were retained such that the rings fragments-once encoded in an FP-were capable of matching individual rings in larger rings systems, especially aromatic rings. Hence, obtained single rings included chemically intact rings, as well as model ring fragments with hybridization states not existing in isolation, representing a special ring feature introduced for the design of CSFP. Fragment Categories For our analysis, we distinguished between core and substituent fragments. For the generation of core fragments, analogue series (ASs) were algorithmically extracted isolated from compounds from medicinal chemistry sources, as specified in the Materials and Methods Section, and their core structures were sampled. Obtaining core structures from ASs ensured that the cores were represented by multiple compounds and were thus generalizable. All extracted cores contained ring structures, and these cores were then decomposed into fused and single rings, as described below. The resulting core fragments were complemented with an equally sized set of most frequently occurring substituent fragments selected from an R-group replacement database we have recently generated and made publicly available [31], resulting from a new methodology for systematically extracting R-groups from compounds [32]. Core Structure Fragmentation Accounting only for the core structure of a compound would produce a single feature, which is of course not suitable for FP design. Therefore, from core structures, all fused and single rings were extracted first by the removal of single bonds between ring systems. Subsequently, fused rings were decomposed into individual ring fragments. All single and fused rings extracted from each core were recorded. Figure 2a-d illustrate the systematic extraction of fused and single rings from cores of exemplary compounds belonging to different ASs (in these cases, removal of all single-bonded substituents from rings yields the core). Importantly, for ring fragments extracted from fused rings, atomic hybridization states were retained such that the rings fragments-once encoded in an FP-were capable of matching individual rings in larger rings systems, especially aromatic rings. Hence, obtained single rings included chemically intact rings, as well as model ring fragments with hybridization states not existing in isolation, representing a special ring feature introduced for the design of CSFP. Ring and Substituent Fragments From the ChEMBL database (version 28) [33], compounds were selected with reported high-confidence activity data for human targets, yielding a total of 67,165 unique active compounds. From these compounds, 7728 unique cores containing at least one ring were obtained. These cores yielded 1116 unique fused and 542 single ring fragments. The R-group replacement database was also extracted from ChEMBL compounds [31]. The benzene ring and H atom were excluded as substituents. Accordingly, the benzene ring was only permitted as a ring fragment, thereby avoiding ambiguous assignments that would occur with high frequency. From the R-group resource, the 500 most frequent substituents obtained by random bond fragmentation and by fragmentation on the basis of retrosynthetic rules [32], respectively, were selected and combined. These subsets displayed a large overlap, as to be expected, and yielded a total of 661 unique substituent fragments. Fingerprint Assembly and Feature Mapping From the isolated ring fragments, the 250 single and the 250 fused rings that most frequently occurred in active compounds from medicinal chemistry were selected and combined with the 500 most frequent R-groups. These fragments were combined to generate CSFP comprising a total of 1000 bits with a balanced composition of rings and substituents, with each fragment assigned to a single bit position, representing a prototypic keyed design. To ensure that test compounds produced bit patterns sufficient for meaningful similarity comparison, substructure relationships between fragments were considered. Hence, individual fused rings or substituents set multiple CSFP bits on if they contained other recorded rings or substituents as substructures, respectively, as illustrated in Figure 1. Hence, as mentioned above, if a fused ring recorded in CSFP was detected, its bit was set on as well as the bits of decomposition fragments, if available. Likewise, if a substituent was detected containing, for example, two others as substructures, three bits were set on (that is, one for the complete substituent and one for each substructure). Compound Activity Classes To evaluate feature distributions in FPs and compare their performance, a set of 30 compound activity classes representing different degrees of difficulty for similarity searching was used. This set included 10 "easy" (structurally homogeneous) classes typically yielding accurate results in similarity searching, 10 "intermediate", and 10 "difficult" (structurally heterogeneous) classes often yielding moderate or low prediction accuracy, as previously reported [34]. Since their original exploration and categorization, many additional compounds have become available for the selected activity classes, which we curated for our study (see Materials and Methods). These up-to-date versions of these activity classes covered diverse targets and contained between 121 and 3159 compounds. Supplementary Materials Table S1 reports the composition of all activity classes. Feature Distribution We first assessed the major goal of CSFP design, that is, producing a reference FP accounting for fewer structural features than the standard 166-bit version of MACCS and the 1024-bit version ECFP4. Therefore, the three FPs were calculated for all compound activity classes and the FP feature distributions were determined. Figure 3 compares these distributions, revealing that the CSFP design goal was fully met. Regardless of the activity class category, the feature distribution was very similar for each FP. However, the number of CSFP features was consistently much smaller than the number of MACCS or ECFP4 features. While MACCS and ECFPs captured a similar number of features, with median values of 55 and 53 features per FP over all combined activity classes, respectively, a corresponding median value of only 28 features per CSFP was observed. Figure 4 shows bit densities of the three FPs for two exemplary compounds, illustrating the lower bit density of CSFP, which was also generally observed. Supplementary Figures S1 and S2 from the Supplementary Materials show the most frequently detected CSFP ring and substituents fragments, respectively, in easy, intermediate, and difficult activity classes, revealing overall balanced distributions of CSFP fragments, with only a few prevalent rings or substituents (such as, for example, the pyridine ring or methyl and amino group). Taken together, these findings confirmed that CSFP captured overall a much lower number of features than the MACCS and ECFP4 standards. Performance Evaluation We then compared the performance of the three FPs in similarity searching and compound classification via machine learning. Since FPs are often used as descriptors in molecular machine learning, we carried out support vector machine (SVM) and random forest (RF) calculations to distinguish compounds from each activity class from a random sample of ChEMBL compounds.. The results of similarity searching and compound classification were evaluated on the basis of different performance metrics (calculation setups and performance measures are detailed in Materials and Methods). densities of the three FPs for two exemplary compounds, illustrating the lower bit density of CSFP, which was also generally observed. Supplementary Figures S1 and S2 from the Supplementary Materials show the most frequently detected CSFP ring and substituents fragments, respectively, in easy, intermediate, and difficult activity classes, revealing overall balanced distributions of CSFP fragments, with only a few prevalent rings or substituents (such as, for example, the pyridine ring or methyl and amino group). Taken together, these findings confirmed that CSFP captured overall a much lower number of features than the MACCS and ECFP4 standards. Performance Evaluation We then compared the performance of the three FPs in similarity searching and compound classification via machine learning. Since FPs are often used as descriptors in molecular machine learning, we carried out support vector machine (SVM) and random forest (RF) calculations to distinguish compounds from each activity class from a random sample of ChEMBL compounds.. The results of similarity searching and compound classification were evaluated on the basis of different performance metrics (calculation setups and performance measures are detailed in Materials and Methods). Performance Evaluation We then compared the performance of the three FPs in similarity searching and compound classification via machine learning. Since FPs are often used as descriptors in molecular machine learning, we carried out support vector machine (SVM) and random forest (RF) calculations to distinguish compounds from each activity class from a random sample of ChEMBL compounds. The results of similarity searching and compound classification were evaluated on the basis of different performance metrics (calculation setups and performance measures are detailed in Materials and Methods). Figure 5 summarizes the results of our systematic similarity search calculations carried out with multiple reference compounds and 1-, 5-, and 10-nearest neighbor (NN) similarity assessment, respectively. As expected, similarity search performance generally decreased from easy over intermediate to difficult activity classes for all FPs, resulting in highly to moderately accurate compound rankings, as assessed on the basis of area under the receiver-operating characteristics curve (AUC ROC) values. From the distributions of obtained AUC ROC values, a clear picture emerged that ECFP4 generally performed best, followed by CSFP, and MACCs. With an increasing degree of difficulty, the performance gaps slightly widened, but the difference between AUR ROC values remained relatively small (mostly falling with a 0.1 value interval). Hence, despite the much lower number of structural features captured by CSFP, its similarity performance exceeded (MACCS) or approached (ECFP4) the accuracy of the standard FPs. All observed differences between AUR ROC distributions were statically significant (Wilcoxon test, p-values < 0.05). Molecules 2022, 26, x FOR PEER REVIEW 7 of 12 Figure 5 summarizes the results of our systematic similarity search calculations carried out with multiple reference compounds and 1-, 5-, and 10-nearest neighbor (NN) similarity assessment, respectively. As expected, similarity search performance generally decreased from easy over intermediate to difficult activity classes for all FPs, resulting in highly to moderately accurate compound rankings, as assessed on the basis of area under the receiver-operating characteristics curve (AUC ROC) values. From the distributions of obtained AUC ROC values, a clear picture emerged that ECFP4 generally performed best, followed by CSFP, and MACCs. With an increasing degree of difficulty, the performance gaps slightly widened, but the difference between AUR ROC values remained relatively small (mostly falling with a 0.1 value interval). Hence, despite the much lower number of structural features captured by CSFP, its similarity performance exceeded (MACCS) or approached (ECFP4) the accuracy of the standard FPs. All observed differences between AUR ROC distributions were statically significant (Wilcoxon test, p-values < 0.05). Compound Classification Similar observations were made for machine learning models used for compound classification. Figure 6 summarizes SVM and RF results for all activity classes based upon balanced accuracy (BA) and the Matthews correlation coefficient (MCC) (Supplementary Figure S3 from the Supplementary Materials shows all results). SVM and RF classification accuracy was, overall, comparably high. For example, even for difficult classes, BA median values greater than 0.95 and MCC median values greater than 0.9 were observed, reflecting accurate calculations. Here, there were small advantages of ECFP4 over CSFP for some but not all performance measures ( Figure S3 from the Supplementary Materials). Differences between MCC value distributions were statistically significant for at least 80% of the activity classes (Wilcoxon test, p-values < 0.05). Overall, however, SVM and RF classification accuracy achieved on the basis of CSFP and ECFP4 was comparable, thus corroborating the results obtained in systematic similarity searching. Compound Classification Similar observations were made for machine learning models used for compound classification. Figure 6 summarizes SVM and RF results for all activity classes based upon balanced accuracy (BA) and the Matthews correlation coefficient (MCC) (Supplementary Figure S3 from the Supplementary Materials shows all results). SVM and RF classification accuracy was, overall, comparably high. For example, even for difficult classes, BA median values greater than 0.95 and MCC median values greater than 0.9 were observed, reflecting accurate calculations. Here, there were small advantages of ECFP4 over CSFP for some but not all performance measures ( Figure S3 from the Supplementary Materials). Differences between MCC value distributions were statistically significant for at least 80% of the activity classes (Wilcoxon test, p-values < 0.05). Overall, however, SVM and RF classification accuracy achieved on the basis of CSFP and ECFP4 was comparable, thus corroborating the results obtained in systematic similarity searching. Based on these criteria, 231,772 unique compounds were obtained with a total of 351,198 activity measurements for 1940 human targets and recorded as SMILES strings [38]. From these high-confidence data sets, a total of 30 compound activity classes (each class represents compounds with activity against a specific target) were selected for benchmarking falling into "easy", "intermediate", and "difficult" categories (10 classes per category) for similarity searching and compound classification, as discussed above. Core Generation and Fragmentation ASs were systematically extracted from all ChEMBL compounds with available highconfidence activity data using the compound-core relationship (CCR) algorithm [39]. The CCR algorithm systematically fragments all combinations of 1-5 exocyclic bonds in a compound (applying a 2:1 core-to-substituent size ratio). From ASs, core structures were extracted and generalized by the addition of hydrogen atoms to all substitution sites [39]. Ring fragments were extracted from core structures using the protocol available in RDKit [35]. Molecular Representations CSFP was compared with the 166-bit version of MACCS and the 1024-bit version of ECFP4 generated using RDKit. Similarity Searching For each similarity search trial, 100 active compounds were randomly selected from each activity class. Then, 10 of these compounds were randomly selected as reference compounds, and the remaining 90 active compounds were added as potential hits to a background database consisting of a random sample of 100,000 ChEMBL compounds Compound Activity Classes From ChEMBL (version 28) [33], compounds with less than 1000 Da, a direct target annotation (target confidence score: 9), and an exact potency value (standard relation: "="), given as K i , IC 50, or K d values, were selected. Interactions with potency values of less than 10 µM or interactions flagged as "inactive", "not active", "inconclusive", or "potential transcription error" were disregarded. After removing undesired targets, such as hERG, serum-albumin, or ABC transporters (i.e., pharmaceutical anti-targets, inhibition of which is undesired), compounds with potential assay interference characteristics were removed using publicly available filters [35][36][37]. Based on these criteria, 231,772 unique compounds were obtained with a total of 351,198 activity measurements for 1940 human targets and recorded as SMILES strings [38]. From these high-confidence data sets, a total of 30 compound activity classes (each class represents compounds with activity against a specific target) were selected for benchmarking falling into "easy", "intermediate", and "difficult" categories (10 classes per category) for similarity searching and compound classification, as discussed above. Core Generation and Fragmentation ASs were systematically extracted from all ChEMBL compounds with available highconfidence activity data using the compound-core relationship (CCR) algorithm [39]. The CCR algorithm systematically fragments all combinations of 1-5 exocyclic bonds in a compound (applying a 2:1 core-to-substituent size ratio). From ASs, core structures were extracted and generalized by the addition of hydrogen atoms to all substitution sites [39]. Ring fragments were extracted from core structures using the protocol available in RDKit [35]. Molecular Representations CSFP was compared with the 166-bit version of MACCS and the 1024-bit version of ECFP4 generated using RDKit. Similarity Searching For each similarity search trial, 100 active compounds were randomly selected from each activity class. Then, 10 of these compounds were randomly selected as reference compounds, and the remaining 90 active compounds were added as potential hits to a background database consisting of a random sample of 100,000 ChEMBL compounds (excluding each activity class). For each class, 20 independent trials were carried out, and the results were averaged. In each trial, the k-nearest neighbor (k-NN) search strategy was applied [40] including 1-NN, 5-NN, and 10-NN calculations. The similarity scores were assessed by calculating the Tanimoto coefficient (Tc) [41]. For 1-NN, a database compound was compared with all 10 reference compounds, and the highest Tc value was selected as the final similarity score. In 5-NN and 10-NN calculations, the top 5 and 10 similarity values were averaged, respectively, to obtain the final similarity score for each database compound. Machine Learning For machine learning calculations, random forest (RF) [42] and support vector machine (SVM) [43] were used. All RF and SVM models were implemented using scikit-learn [44]. Random Forest The RF is a supervised machine learning algorithm that derives an ensemble of decision trees generated from randomly selected training instances using bootstrapping. In predictions, each tree yields a class label for a test instance, and the final class label is determined by an ensemble majority vote [42]. Support Vector Machine SVM is a supervised learning method deriving a hyperplane in feature space that optimally separates training instances with different class labels by maximizing the separating margin of the hyperplane. If linear separation is not possible in the feature space, a kernel function is applied to project the training data into a higher-dimensional space where linear separation might become possible [43]. Model Building and Hyperparameter Optimization For each activity class, 50% of the compounds were randomly selected for training and hyperparameter optimization. The remaining 50% of the compounds served as an external validation set. For hyperparameter optimization, 10-fold internal cross-validation and grid search were applied. Optimal hyperparameters were chosen based on the mean balanced accuracy across all trials. For RF, the number of trees was determined by testing 25, 50, 100, 200, and 400 trees, and the minimum number of samples required to split an internal node by testing 2, 3, 5, and 10 samples. For all remaining hyperparameters, default settings were used. For SVM, the cost hyperparameter C, which regulates the trade-off between misclassified samples and the margin size, was optimized using candidate values of 0.1, 1, 10, 50, 100, 200, and 1000. SVM models were derived using the Tanimoto kernel, a preferred choice for binary fingerprints [45]. Class weights were set to "balanced". For all remaining hyperparameters, default settings were used. Predictions RF and SVM models were applied to predict 50% of the compounds from each activity class not used for training (positive instances). As negative instances, three times the number of positive instances were randomly selected from ChEMBL (excluding each activity class). For each class, 20 independent trials were carried out, and the results of the predictions were averaged. Performance Measures The performance of RF and SVM models was evaluated on the basis of different measures including balanced accuracy (BA) [46], Matthew's correlation coefficient (MCC) [47], F1 score [48], precision, and recall. Similarity searching performance was evaluated according to the area under the ROC curve (AUC ROC) [49]. precision = TP TP + FP (4) recall = TP TP + FN (5) where TP, TN, FP, and FN stand for true positives, true negatives, false positives, and false negatives, respectively. For results of similarity searching and compound classification, statistical significance assessment was based on AUC ROC and MCC values, respectively, using the nonparametric Wilcoxon test [50]. Conclusions In this study, we have introduced a new generally applicable 2D FP designed to investigate the question of whether or not fewer structural features than commonly captured in state-of-the-art 2D FPs might be sufficient for correctly detecting molecular similarity relationships in similarity searching and compound classification. CSFP was assembled from ring and substituent fragments systematically extracted from biologically active compounds. A key aspect of its design is separately accounting for substructure relationships between rings and substituents, hence yielding multiple bit settings for fused rings and subsets of larger substituents and ensuring the presence of minimally required bit density for meaningful FP comparison. CSFP was shown to contain significantly fewer structural features than MACCS or ECFP4 but exceeded the predictive performance of MACCS in similarity searching and machine learning and approached (or met) the performance of ECFP4. Taken together, these findings demonstrated that a smaller number of FP features than that currently used is sufficient for the accurate detection of compound similarity relationships indicative of similar biological activity. Although CSFP was primarily designed as a research tool, its chemically intuitive nature and high-performance level also render it favorable for practical applications in medicinal chemistry. On the basis of the computational protocols provided herein and the substituent resource we have made publicly available, the CSFP design can be easily reproduced, modified, and further extended. Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules27072331/s1; Figure S1, Most frequent ring fragments, reports the CSFP frequency of ring fragments from easy, intermediate, and difficult activity classes, respectively. Figure S2, frequent substituent fragments, reports the CSFP frequency of substituent fragments from easy, intermediate, and difficult activity classes, respectively. Figure S3, Compound classification. Boxplots report RF and SVM results for MACCS, CSFP, and ECFP4 on the basis of different performance measures (see Materials and Methods) across all activity classes. Table S1. Compound activity classes. For each of the 30 activity classes used for our analysis, the ChEMBL target ID and the number of compounds are reported. Data Availability Statement: All calculations were carried out using publicly available data.	2022-04-06T15:09:21.875Z	2022-04-01T00:00:00.000	{ "year": 2022, "sha1": "178ae6a33dc1a3af28a8e55da7e8b522cfec21ab", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1420-3049/27/7/2331/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "b993f92d9598419f1473da75c2080e98aaf8c568", "s2fieldsofstudy": [ "Chemistry" ], "extfieldsofstudy": [ "Medicine" ] }
199442489	pes2o/s2orc	v3-fos-license	Estimating Unobserved Individual Heterogeneity Using Pairwise Comparisons We propose a new method for studying environments with unobserved individual heterogeneity. Based on model-implied pairwise inequalities, the method classifies individuals in the sample into groups defined by discrete unobserved heterogeneity with unknown support. We establish conditions under which the groups are identified and consistently estimated through our method. We show that the method performs well in finite samples through Monte Carlo simulation. We then apply the method to estimate a model of low-price procurement auctions with unobserved bidder heterogeneity, using data from the California highway procurement market. Introduction The empirical analysis of many economic settings requires accounting for unobserved individual heterogeneity (UIH) which reflects agent-specific factors that influence agents' decisions but are not recorded in the data. Failing to account for UIH generally leads to biased estimates and affects the validity of counterfactual prediction. In this paper, we consider a generic economic model where UIH induces a group structure among agents according to their types. We provide conditions for identification of the group structure, and propose a method to recover the group structure from data. Our main idea is based on the insight that UIH often implies pairwise inequality restrictions on endogenous observable quantities. For instance, in multi-attribute auctions, bidders with higher (unobserved) quality levels have a greater chance of winning the auction, controlling for the bid and the set of competitors. In a labor market setting, agents with higher (unobserved) productivity receive higher wages than less productive ones. In Section 2.2 we provide further examples of economic applications in which pairwise inequality restrictions arise naturally from the behavior of agents in equilibrium. We develop a statistical method to recover the group structure (that is, to classify individuals into groups defined by UIH) using a pairwise comparison approach. Our method treats UIH as individual-specific discrete parameters which may affect the distribution of other observed or unobserved variables. Such flexibility is important in structural models where individuals interact strategically and the UIH of all agents jointly affects the equilibrium outcome. Our method is nonparametric, i.e., does not assume that UIH belongs to any parametric family of distributions, or that the support of discrete UIH is known. A naive classification of individuals into groups based on pairwise inequality tests does not necessarily yield a coherent group structure in finite samples in general. Our method recovers the whole group structure by sequentially sub-dividing the set of agents on the basis of the p-values of tests of pairwise inequality restrictions. The method recovers the group structure for each assumed number of groups, and then selects the number of groups (and the associated group structure) using a penalization scheme. We show that our estimator of the group structure is consistent under mild regularity conditions and performs well in small samples. In many settings, classifying individuals into groups defined by UIH offers key economic insights. For example, our method can be used to identify colluding bidders in auctions, and firms' cost asymmetries or product quality differences. In addition, recovering the group structure also often serves as the first step for estimating structural models with strategic interactions, such as dynamic industry models or auctions with asymmetric bidders. 1 This approach offers a feasible way to identify and estimate games with UIH. Specifically, a traditional approach in a setting with UIH would be to treat UIH as "fixed effects" 1 Estimation of discrete unobserved individual heterogeneity does not affect subsequent estimation of other structural parameters in terms of pointwise asymptotics. However, establishing uniform asymptotics remains an open question. This problem is analogous to that of post-model-selection inference. For discussion on the issues, see Pötcher (1991), Leeb and Pötcher (2005), and Andrews and Guggenberger (2009) and the references therein. Uniform asymptotics in our setup is complex because of the need to consider every possible direction of local perturbation from the actual group structure in data-generating process. A full theoretical investigation of the issue in our context merits a separate paper. and estimate them jointly with other structural parameters. This approach poses practical challenges in a setting with agent interdependence, especially when equilibrium outcomes admits no closed-form expressions. First, it is generally not obvious what variation in the data may identify model components including the fixed effects in such settings. One of the contributions of our paper is to point out the variation which identifies the group structure associated with "fixed effects." Further, we propose to separate the recovery of the group structure from the estimation of other structural parameters. Recovering the group membership of every agent facilitates, and is often needed for, identifying the remaining structural elements in models with strategic interactions. A classical example is English auctions among bidders with unobserved types (in the sense that bidders' private values are drawn independently from the distributions "labeled" by bidder type) and where the data only report the transaction price and the identity of the winner. Athey and Haile (2007) show that the distributions of private values cannot be identified in this model if the type of the auction winner is unknown. We provide details and additional examples in the supplemental note to this paper. Finally, our approach offers a way to estimate games with UIH with substantially low computational cost, compared with the alternative approach of estimating the fixed effects jointly with other structural parameters. For example, consider an environment with a large number of players where many independent games (each containing only a small subset of players) are observed in the data. The numerical optimization (such as simulated GMM or MLE) requires evaluating the objective function which involves computing equilibrium for each value of the fixed effects and other structural parameters. This can be computationally prohibitive in practice. In contrast, our method allows the game to be solved only for the estimated configuration of the agents' group memberships rather than for every possible configuration as is required under the joint estimation. 2 Our method is advantageous especially in settings where the number of agents is moderately large but each market (observation) in the data involves only a small subset of participants. For example, the total number of participants may be several hundreds but each market may contain only several participants. In this case, despite the large number of markets observed, the researcher may have only a small number of markets which contain the same set of participants. We call this issue the problem of the sparsely common set of agents. In such settings, the researcher cannot build inference on the conditional moments given the full set of participating agents in a market, as typically done in the structural empirical literature, because we do not have many such observations. Hence 2 Krasnokutskaya, Song, and Tang (2018) used this classification method as a first step in the structural analysis of an online service market for computer coding. the researcher needs to "aggregate" the markets or the agents in order to conduct reliable inference with sufficiently many observations. Pairwise restrictions are testable with accuracy even when the data exhibit sparsely common sets of agents since the number of markets where a given pair of agents is present tends to be large even if the number of markets with the same set of participants is small. Thus, pairwise restrictions and the classification procedure offer a natural way to aggregate agents into groups which permits estimation of other primitives. We investigate the finite sample performance of our classification method in Monte Carlo simulation. The data-generating process (DGP) is a lowest-price procurement auction among asymmetric bidders whose independent private values are drawn from distributions with different means. We report the outcome of classification for DGPs with various numbers of bidders and group structures. Our classification method works well. Its performance is better when the number of bidders and groups are smaller relative to the number of observed markets/games, and when the differences between groups are larger. We also find that the impact of classification errors on subsequent estimation of other structural parameters in the game is non-substantial. We analyze the California highway procurement market by applying our classification method to a model of asymmetric lowest-price procurement auction. Existing empirical studies of auction markets typically emphasized asymmetry in bidders' private values associated with their observable characteristics. 3 In comparison, we allow the bidders' private values to be drawn from heterogeneous distributions with different means. To account for other sources of cost heterogeneity, we control for bidders' distance to the project site. We also accommodate possible endogeneity in the competitive structure. We use the classification method to recover the unknown group structure (i.e., partition bidders into groups with different mean costs). Then, using this estimated group structure, we estimate group-specific cost distributions using GMM. Our estimates indicate that the bidders in the data come from several unobserved groups with substantial differences in mean costs. We also find that ignoring such unobserved bidder heterogeneity would lead to biased estimates of how bidders' costs depend on various factors. Related Literature. One of the popular methods of accounting for UIH in structural modeling is to adopt finite mixtures. (See Hu (2008), Hu and Schennach (2008), Kasahara andShimotsu (2009), Hu andShum (2012), Hu and Shiu (2013), Hu,McAdams,3 For example, Athey, Levin, and Seira (2011), Roberts and Sweeting (2013) and Aradillas-Lopez, Gandhi, and Quint (2013) accounted for the bidder heterogeneity associated with size in the timber market ('mills' vs 'loggers'); Krasnokutskaya and Seim (2011), Jofre-Bonet and Pesendorfer (2003) and Gentry, Komarova, and Shiraldi (2016) incorporated bidder participation differences in highway procurement market ('regular' vs 'fringe' bidders); Conley and Decarolis (2016), Asker (2010) and Pesendorfer (2000) allowed for bidder heterogeneity in collusive behaviors. and Shum (2013), and Henry, Kitamura, and Salanie (2014). See also See Kasahara and Shimotsu (2014) and Kasahara and Shimotsu (2015) for estimating and testing for the number of mixture components in finite mixture models.) The finite mixture modeling assumes that the UIH is a random variable drawn from some unknown distribution. The goal is to identify this distribution and estimate it from data. It does not require each individual to appear in many independent games. In contrast, our approach aims to classify individual agents in the sample into disjoint groups defined by their realized unobserved types, using their participation outcomes in many games. Thus the two approaches are fundamentally different both in their aims and their data requirements. While general identification results have been developed in this literature of finite mixture models (see, e.g., Bonhomme, Jochmans, and Robin (2016)), implementation of the finite-mixture method is impractical in our set-up due to the issue of sparse commonality, and technical issues associated with high dimensionality. 4 The classification algorithm we propose is related to the clustering method in statistics. (See, e.g., Chapter 14 of Hastie, Tibshirani, and Friedman (2009).) The main difference is that the clustering method groups individuals based on similarity of their observed attributes, whereas our approach groups individuals based on their unobserved attributes. To do so, we exploit the relationship between endogenous outcome and the unobserved types of individuals implied by an economic model. Our method also requires a data structure different from clustering methods. The literature of clustering methods mostly considers a set-up in which each cross-sectional unit is observed once, whereas our method uses many observations per individual in the sample. Also related to our approach is the literature of panel models with group level heterogeneity. For example, Sun (2005) introduced a linear panel model where parameters take values in a finite set according to a logistic probability, and offered methods of estimating the group structure. Song (2005) considered a panel model with finite-valued nonstochastic parameters and produced an algorithm to recover the unobserved group structure in large panel models. Lin and Ng (2012) provided a method of estimating a panel model using threshold variables when the group membership is unknown. Su, Shi, and Phillips (2016) developed a new Lasso method to recover the unknown groupspecific parameters. Bonhomme and Manresa (2014) proposed a k-means clustering algorithm to recover the group structure in a linear panel model. These papers often focus on models which admit a reduced form for the dependent variable in which its functional relation to UIH is made explicit. In contrast, our method targets a set-up where the dependence of the outcome variables on the UIH arises implicitly through equilibrium 4 When each market is drawn from a finte mixture distribution, and there are I agents with each having a type from S values, the number of the mixture components becomes S I which can be very large in pratice, even when I is a moderate number such as five or ten. contraints in games, and the group structure of UIH is identified only through pairwise inequality restrictions. Thus, the approaches developed in the panel literature are not applicable in settings our proposal focuses on. Roadmap. This paper is organized as follows. Section 2 introduces the basic environment and defines pairwise inequality restrictions. This section also provides several examples from various contexts to motivate our classification method. Section 3 establishes identification of the unobserved group structure using pairwise inequality restrictions. Section 4 proposes a consistent estimator of the group structure. Section 5 provides results from Monte Carlo simulation. Section 6 presents the empirical application. Section 7 concludes. Further examples and mathematical proofs are provided in the supplemental note of this paper. The note also contains further simulation results and details in our empirical application. Pairwise Inequalities in Game Models We consider a setting where the econometrician observes L games, and in each game, a set of agents interact with each other. Each agent i is associated with a non-stochastic type, q i , which is not observed by the researcher. We assume that the type is finte-valued so that q i ∈ Q 0 = {q 1 , ...,q K 0 }, withq 1 < · · · <q K 0 . This induces an (ordered) partition (N 1 , N 2 , ..., N K 0 ) of the set N of agents such that for each k = 2, ..., K 0 , N k consists of agents with higher type than those in N k−1 . The group structure is characterized by a function τ : N → {1, ..., K 0 } that links the identity of a player to his unobserved type so that q i =q τ (i) and for each k = 1, ..., K 0 , The data available to the researcher contain for every observation l = 1, ..., L: a vector of observable characteristics for all the agents involved, {X j,l } j∈S l , as well as at least one but possibly multiple vectors of outcome variables, Y l = {Y j,l } j∈S l , where S l is the set of players involved in each observation (e.g., a game or a market). Our main focus is on recovering the ordered partition (N 1 , N 2 , ..., N K 0 ) of agents from data. The main insight of our paper begins with the observation that in many structural models, the ordering among q i 's (or equivalently τ (i)'s) coincides with the ordering between indexes that can be estimated consistently. Specifically, we use pairwise indexes δ ij and δ 0 ij that satisfy the following relations: 5 where the indexes δ ij and δ 0 ij can be consistently estimated using the sample. In many applications, we can take the indexes as (a variant of) the following form: where F is a known distribution or the distribution of an observable random vector. For example, suppose that the outcome Y i,l admits the following reduced form: where g is a function that is strictly increasing in τ (i) and η i,l is an unobserved component that is independent of X i,l . Then under regularity conditions, we obtain the pairwise relations (2.1) with (2.2). The main advantage of our approach is that we do not require an explicit characterization of the reduced form g. Due to this flexibility, our approach is most useful for analyzing UIH in structural models where the reduced form for outcomes arises only implicitly through equilibrium constraints. In such a setting, the sign of the indexes δ ij represents the pairwise relation which says that between any two agents, one agent's type is higher than the other if and only if his outcome tends to be higher than that of the other. As we demonstrate through examples below, many structural models imply these pairwise relations through indexes δ ij and δ 0 ij . The main goal of this paper is to develop a statistical procedure to recover the group structure τ from data. Our method relies only on the pairwise inequality restrictions in (2.1). Thus so far as the group structure is concerned, the pairwise comparison indexes δ ij and δ 0 ij play the role of a sufficient statistic; the recovery of the group structure does not rely on other details of the structural model. Examples We now provide examples of how pairwise inequality restrictions arise as equilibrium implications in a variety of commonly studied empirical contexts. 5 For the sake of concreteness, our exposition in the paper focuses on this form of indexes. Our procedures rely on the indexes only through the availability of consistent tests of pairwise inequality restrictions: δ ij > 0 and δ 0 ij = 0. As long as such consistent tests are available, one can use our method for other forms of pairwise indexes. 2.2.1. Unobserved Quality in Multi-attribute Auctions. Consider a simplified version of multi-attribute auctions in Krasnokutskaya, Song, and Tang (2018) that abstracts away from observed auction and seller heterogeneity. Let N denote the population of sellers and S l the set of sellers who submitted bids for a project l. Each seller has a discrete unobservable quality: q i ∈ {q 1 , ...,q K 0 }, withq k <q k whenever k < k . Such a quality is known to buyers but not reported in data. The buyer for project l selects a seller among those who submitted bids or chooses an outside option to maximize his payoff. The payoff to the buyer from engaging services of seller i ∈ S l is given by U i,l = α l q i + i,l − B i,l whereas the payoff from an outside option is U 0,l . Here α l is a non-negative weight the buyer gives to the seller's quality relative to the seller's bid, whereas i,l reflects a buyerseller match-specific stochastic component. Let us suppress the auction subscript l and define for any two sellers i, j, for all b on the intersection of the supports of B i and B j . Suppose α, S, {C i , i } i∈S are mutually independent. 6 Proposition 1 of Krasnokutskaya, Song, and Tang (2018) showed that for any b on the intersection of bid supports. On the basis of this property the comparison indexes can be constructed as follows: Note that the comparison indexes do not depend on other details of the structural model such as specific parametric assumptions for the distribution of buyers' tastes. Firms' Cost Efficiency and Pricing Decisions. Consider a population of n firms or brands, each of which produces a single brand of product. The data consists of independent markets indexed by l = 1, ..., L. The marginal cost for firm i on market l is c i,l = ϕ(w i,l , q i , η i,l ), where w i,l are observable cost shifters, q i a brand-specific unobserved heterogeneity that is fixed across markets, and η i,l 's are i.i.d. idiosyncratic noises independent from w i,l and q i . We may interpret q i as a measure of firm i's cost efficiency. Firms have complete information about each others' cost efficiencies. 7 Firms in the population are partitioned into groups with different levels of Let σ i,l (x l , p l , Ω l ) denote firm is' market shares, which is a function of product attributes (x l = {x i,l } i∈S l , where S l denotes the set of brands in market l) and prices (p l = {p i,l } i∈S l ) 6 This holds, for example, if sellers are not informed of the weights or outside option of the buyer, or the identities of other sellers in S l . 7 This assumption is plausible in certain industries where production efficiency is mostly determined by firms' technology or equipment that is publicly observable. conditional on the set of products available in market l and other market factors denoted by Ω l . The profit for firm i in market l is: π i,l = (p i,l − c i,l )σ i,l (x l , p l )M l , where M l is a measure of potential consumers in market l. In any pricing equilibrium with an interior solution, the first-order condition implies Notice that if η i,l is independent from q i and w i,l , and ϕ(w i,l , q i , η i,l ) is strictly monotone in q i then so is the right-hand side of (2.4), which can be constructed from estimates of the demand system. Hence, for any pair i, j ∈ N , , for all w 0 , where z i,l is defined as the quantity on the right hand side of (2.4). The statement is also true when both inequalities are strict. Thus we can define a pairwise comparison index where F is the distribution of w i,l . In equilibrium δ ij > 0 if and only if q i > q j . Likewise, define δ 0 ij by replacing max{·, 0} in the integral in δ ij with the absolute value. These pairwise comparison indexes do not condition on specific identities of firms in a market. Assortative Matching in Labor Market. Sorting of heterogeneous employees across heterogeneous firms has been studied in Lentz and Mortensen (2010), Abowd, Kramarz, andMargolis (1999), andLise, Meghir, andRobin (2011). In a typical setting, firms are heterogeneous in the productivity from a given worker ceteris paribus. Workers differ in their unobservable ability q i . Under further restrictions (see Eeckhout andKircher (2011) andHagedorn, Law, andManovskii (2016)), workers with higher ability would in equilibrium earn higher wages than co-workers at the same firm, holding other things equal. This forms a basis for pairwise comparisons. Specifically, let w i,f,t = W (q i , X i,t , Ω f,t ) denote the wage worker i earns at time t while employed by firm f , where W is a nonstochastic function. Here Ω f,t captures all the relevant firm-specific unobservable factors while X i,t reflects worker i's characteristics other than q i . Using N f,t to denote the set of workers employed by firm f at time t, we define the comparison index as where F is the distribution of X i,t . Then δ ij > 0 if and only if q i > q j , under regularity conditions such as strict mononicity of W in q i . Likewise before, define δ 0 ij by replacing the max operator in δ ij with its absolute value. In this setting comparison of workers is complicated by the (unobserved) firm heterogeneity and sorting of workers across FIGURE 1. The first panel shows an example of a data set where the set of participants is the same across all markets. The second panel shows another example of data where only a small fraction of markets share the same set of participants. As illustrated here, there are only three markets where the set of participants is precisely {1, 3, 4}. The third panel shows that there are many more markets in the second example where both agents 1 and 3 (represented by white ellipses) participate. Thus a researcher can estimate population quantities that condition on the joint participation of these two agents with better accuracy than quantities that condition on the whole set of participants. firms. Pairwise comparisons allow researchers to circumvent these issues by focusing on workers' wages earned while they are employed by the same firm. Sparsely Common Set of Agents and Pairwise Inequalities Our pairwise comparison method is most useful in a setting where players appear in a market only sparsely. To express this data feature, define for any S ⊂ N , Thus L(S) represents the set of markets where the set of participants in a market S l is precisely S. In this paper, we refer to the setting as that of a sparsely common set of agents, if the proportion max S⊂N \|L(S)\|/L is negligible in finite sample. In other words, only a small fraction of the markets in the sample share exactly the same set of participants. The setting with a sparsely common set of agents is illustrated in Figure 1, where each column symbolizes a "market" and each row an individual agent. The ellipses in each column represent agents participating in a market. The first panel shows a standard setup where all the agents appear in all the markets. The second panel shows an example of a data set where only very few markets share exactly the same set of participants {1, 3, 4}. Therefore, the conditional choice probability given the same set of agents simultaneously participating in the market cannot be accurately estimated. However, if we focus on only subsets with two agents {1, 3}, there are many more markets in which the two agents participate. Aggregating over these markets, one may infer accurately the ordering between the two agents using an inequality test. Given the p-values from inequality tests across pairs of agents, it remains to recover the whole group structure of the agents from these pairwise p-values. We develop an algorithm that recover the group structure from the pairwise p-values consistently. Identification of the Ordered Group Structure We say that agents (i, j) are comparable if there exist consistently estimable pairwise indexes δ ij and δ 0 ij such that (2.1) holds. In this identification analysis, we assume that a researcher knows whether each pair of agents is comparable through some pairwise comparison index or not. The determination of such comparability can be done in practice by checking whether the data contains sufficiently many markets which allow for reliable estimation of the pairwise indexes. Let E be the collection of pairs (i, j) that are comparable. We refer to comparable agents as adjacent (or linked), so that the set E forms the set of edges in a graph on the set of agents N . We call this graph (denoted by G = (N, E)) the comparability graph. 8 We say a group structure τ is identified if it is uniquely determined once the comparability graph G and the vectors of pairwise indexes (δ ij , δ 0 ij ) ij∈E are known. Let us explore the identification of τ given the comparability graph G and the vector of pairwise indexes. It is easy to see that if E contains only a small subset of possible pairs, we may not be able to identify the group structure. The identification of the ordered group structure τ is not guaranteed even when many pairs of agents are comparable. For example, even if G is a connected graph (where any two agents are connected at least indirectly), the ordered group structure τ may not be identified. This is illustrated in a counterexample in Figure 2. Certainly, when every pair of agents are adjacent in the graph G, i.e., G is a complete graph, the ordered group structure τ is identified. 9 8 In a graph (or network) G = (N, E) the set N represents the set of vertices (or nodes) and E consists of some pairs ij, with i, j ∈ N , where each pair ij is called an edge (or link). If (i, j) ∈ E, we say that i and j are adjacent. A path is a set of vertices {i 1 , i 2 , ..., i M } such that i 1 i 2 , i 2 i 3 , ...i M −1 i M ∈ E. Two vertices are called connected if there is a path having i and j as end vertices. A graph is called connected if all pairs of vertices are connected in the graph. 9 If all pairs of agents are comparable, we can split the set of agents into one group with the lowest type and the other group with the remaining agents. Then we split these remaining agents into one group with is violated. The first panel depicts the comparability graph as one connecting 6 vertices (or nodes). The second panel shows the τ -collapsed graph where the two comparable nodes 2 and 3 that have the same type are collapsed into one node named 23. The last panel shows that Nodes 23, 4, and 5 (expressed as solid black nodes) are identified, because they are on a monotone path of length K 0 − 1 = 2. In this example, Nodes 1 and 6 are not identified and thus the comparable graph does not lead to the identification of the group structure. Below we establish a necessary and sufficient condition for the group structure to be identified from an incomplete graph G and the pairwise comparison indexes. Let us introduce some definitions. the lowest type within these agents and the remaining agents. By continuing this process, we can identify the whole group structure. Definition 3.1. (i) A graph G τ is the τ -collapsed graph of G if (a) any two adjacent vertices i and j in G with τ (i) = τ (j) collapse to a single vertex (denoted by (ij)) in G τ , (b) any edge in G joining a vertex k to either i or j joins vertex k to (ij) in G τ and (c) all the remaining vertices and edges in G τ consist of the remaining vertices and edges in G. The τ -collapsed graph of G is constructed by reducing any comparable pair of agents in G who have the same type to a single "agent", and retaining edges as in the original graph of G. Certainly, a τ -collapsed graph G τ is uniquely determined by δ 0 ij 's and G. Any pair of adjacent agents in the τ -collapsed graph must have different types, and hence the types of agents on a monotone path are strictly monotone. This means that every vertex on a monotone path in G τ of length K 0 − 1 is identified. 10 Also by similar logic, every vertex on a monotone path with end vertices i H and i L is identified if the path has length τ (i H ) − τ (i L ) and the end vertices i H and i L are identified. Using these two facts, we can recover the set of vertices that are identified as follows. First, let N [1] ⊂ N denote the set of vertices such that each vertex in the τ -collapsed graph G τ is on a monotone path in G τ of length K 0 − 1. For j ≥ 1 generally, let N [j+1] be the set of vertices each of which belongs to a monotone path, say, P , such that its end vertices i H and i L are from N [j] and τ (i H ) − τ (i L ) is equal to the length of the monotone path P . Then define Given G τ , N * is uniquely determined as a subset of N . It is not hard to see that if N = N * and K 0 is identified, the type structure τ is identified. The following theorem shows that this condition is in fact necessary for the identification of τ as well. Theorem 3.1. Let G be a given comparability graph and G τ be its τ -collapsed graph. The group structure τ is identified if and only if there exists a monotone path in G τ whose length is equal to K 0 − 1 and N = N * . No monotone path in G τ can have length greater than K 0 − 1. Note that there exists a monotone path in G τ whose length is equal to K 0 − 1 if and only if K 0 is identified. The conditions in the theorem are obviously satisfied if G contains a monotone path that is monotone and covers all the vetrices. The latter condition is trivially satisfied when G is a complete graph. Figure 3 gives a counterexample where the condition that there exists a monotone path in G τ whose length is equal to K 0 − 1 is satisfied, but N = N * so that the comparability graph does not lead to the identification of the group structure. Pairwise Hypothesis Testing Problems In this section, we develop a method to estimate the group structure consistently for the case where the comparability graph is complete, so that we take E to be all ij with i, j ∈ N, i = j. We first formulate three pairwise hypothesis testing problems for each comparable pair ij ∈ E: In most examples, we have various tests available. Instead of committing ourselves to a particular method of hypothesis testing, let us assume generally that we are given p- respectively. Let L be the size of the sample (i.e., the number of the markets or games) that is used to construct these p-values. We will explain conditions for the p-values later and explain details for construction of p-values using bootstrap in Section 4.3. The Classification Method Our classification method consists of two generic algorithmic components: the Split Algorithm and the Selection-Split Algorithm which contains the Split Algorithm as a component. 4.2.1. The Selection-Split Algorithm. Let us introduce a method of obtaining an ordered partition (N 1 ,N 2 ) of a given set N using p-valuesp s ij , s ∈ {+, 0, −}. Definition 4.1. For a subset N ⊂ N , we say that the ordered partition of N into (N 1 ,N 2 ) is obtained by the Split Algorithm if it is obtained as follows. For each i ∈ N , we let where r L → ∞ satisfies Assumption 4.1 below. 11 Set i * = argmin i∈N min{s 1 (i), s 2 (i)}, where 11 In many cases, it suffices to consider a sequence such that r L / log L → 0. In practice, we propose r L = (log L) 1/3 which satisfies Assumption 4.1 below under lower level regularity conditions. See Section C.3 in the supplemental note for details. From our simulations, we find that even choosing r L = 0 works well in finite samples. (We set s 1 (i) = 0 ifN 1 (i) is empty, and similarly with s(i).) Then we take The setN 1 (i) estimates the set of agents of lower type than i, and the setN 2 (i) estimates the set of agents of higher type than i. Let A necessary condition forN 1 (i) to coincide with N 1 (i) is that i has higher type than those inN 1 (i). The more negative the quantity s 1 (i) is, the more likely that this necessary condition is met. A similar observation applies to s 2 (i) as well. Thus we choose a partition that minimizes min{s 1 (i), s 2 (i)} over i. Suppose that we are given an ordered partition (N 1 , ...,N s ) of N . The Selection-Split Algorithm that we propose produces an ordered partition (N 1 , ...,N s+1 ) of N from (N 1 , ...,N s ) using two steps, the Selection Step and the Split Step, as follows. The Selection Step chooses a groupN k * that is most likely to contain agents with heterogeneous types and the Split Step splits this group into two sets using the Split Algorithm. The Selection-Split algorithm depends on the data only through the p-valueŝ p s ij , s ∈ {+, 0, −}. The Classification Method. For a given positive integer 2 ) using the Split Algorithm to N , and apply the Selection-Split Algorithm sequentially to obtain (N K ) for a given number K. For each K, we definê where g(L) is slowly increasing in L. 12 We takeTK = (N K ) to be our estimated group structure. The componentV (K) measures the goodness-of-fit of the classification, and the second component Kg(L) represents a penalty term that prevents overfitting. We show that TK is consistent for the underlying group structure T under regularity conditions. Constructing p-Values Using Bootstrap In most applications, we can use bootstrap to construct p-values for testing the inequality restrictions of (4.1). 13 For the sake of concreteness, we explain the bootstrap procedure along the proposal made by Lee, Song, and Whang (2018). Suppose that we are given observations denotes the observations pertaining to market l and Z i,l denotes the vector of observations specific to agent i. Suppose that for each pair of agents i and j, there exists a nonparametric function, say, To construct a test statistic, we first estimate r ij (x) using the sample {Z l } L l=1 to obtainr ij (x) (e.g., using a kernel regression estimator). Then we construct the following indexes:δ For p-values, we re-sample {Z * l } L l=1 (with replacement) from the empirical distribution of {Z l } L l=1 and construct a nonparametric estimatorr * ij (x) for each pair (i, j) in the same way as we did using the original sample. Using these bootstrap estimators, we construct the following bootstrap test statistics: Note that the bootstrap test statistic involves recentering to impose the null hypothesis. Now, the p-values,p + ij ,p − ij , andp 0 ij can be constructed from the bootstrap distributions of δ * ij ,δ * ji , andδ 0 * ij respectively, usingδ ij ,δ ji andδ 0 ij as test statistics. Consistency of Classification We prove consistency of the estimated classificationTK as L → ∞ while n fixed. (The consistency results and the proof (with high level conditions) for the case of both n and L increase to infinity can be found in the supplemental note.) Let P be the collection of 12 The choice of g(L) = log log L appears to work very well from our numerous Monte Carlo simulation experiments. 13 See Bugni (2010), Andrews and Shi (2013), Chernozhukov, Lee, and Rosen (2013), Lee, Song, and Whang (2013), and Lee, Song, and Whang (2018), among many others, and references therein. the distributions P of the whole vector of the observations in each market l. For each ε > 0, ij ∈ E and s ∈ {+, 0, −}, we define where we write the pairwise indexes δ s ij as δ s ij (P ) to reflect that the pairwise indexes depend on P . Thus P s 0,ij is the collection of probabilities under the pairwise null hypothesis H s 0,ij , and P s ε,ij is the collection of probabilities under the pairwise alternative hypotheses H s 1,ij such that δ s ij (P ) is away from zero at least by ε. Then we define We assume that the p-value takes the following form: whereF s ij is a CDF andT s ij is a random variable both of which depend on the data. Typically,T s ij represents an appropriately normalized test statistic andF s ij represents the CDF of the bootstrap distribution of the test statistic after recentering. We make the following assumption. Assumption 4.1. There exist sequences λ L → ∞ and ρ L → 0 and constants c s ij , s ∈ {+, 0, −} such that along each sequence of probabilities P L ∈ P 0,ε and for each pair i, j ∈ N , the following holds for all s ∈ {+, 0, −}, as L → ∞. ( Assumption 4.1 is typically satisfied for various choices of test statistics that arise in the literature of moment inequality testing. See Section C.2 of the supplemental note for some lower level conditions for the case of testing procedures based on Lee, Song, and Whang (2018). In this case, F s ij,∞ is a standard normal CDF. Theorem 4.1. Suppose that Assumption 4.1 holds, and that g(L) → ∞ and g(L)/r L → 0 as L → ∞. Then, for any ε > 0, along a sequence of probabilities P L from P 0,ε , and the estimated group structureTK satisfies that as L → ∞, Note: n denotes the total number of the bidders; K 0 denotes the number of the groups; n k denotes the number of actual bidders from group k. For each structure in the simulation design, groups all have the same number of bidders. The proof of Theorem 4.1 proceeds in two steps. First, we show thatT K 0 is consistent for T . Second, we show thatK is consistent for K 0 . To see the intuition for this second step, note that when K ≥ K 0 , the componentV (K) is O P (1), and when K < K 0 , the componentV (K) diverges at a rate faster than g(L). From this, we obtain thatK is consistent for K 0 . Finite Sample Performance of the Classification We use a model of a first-price procurement auction with asymmetric independent private costs to study performance of our classification procedure. (See Appendix B.1 of the supplemental note.) Bidders are classified into K 0 groups. We abstract away from the formation of equilibrium strategies, and draw bids from a normal distribution N (µ k , σ 2 ). Let L denote the number of auctions in which any given pair of bidders participates. We consider two specifications of µ k 's. In one specification, µ 1 = 2.0, µ 2 = 2.6, µ 3 = 3.2, and µ 4 = 3.8 with increment D µ = 0.6, and in the other specification, µ 1 = 2.0, µ 2 = 2.2, µ 3 = 2.4, and µ 4 = 2.6 with increment D µ = 0.2. The variance σ 2 is taken to be 0.25. Table 1 summarizes the designs of group structures in our simulation. The first two structures involve a total of 12 bidders and the last two 40 bidders. The first and third are designed to be coarser group structures than the second and fourth respectively. We construct p-values using the procedure in Section 4.3 and obtain group classification from 500 simulated samples. For each estimate, we used 200 bootstrap iterations to calculate p-values. To evaluate the performance of our classification method, we define a measure of discrepancy between two ordered partitions T 1 and T 2 : Note: n is the number of bidders in data; and L the number of markets.K 0 is the average number of estimated groups in 500 simulation samples. EAD is the average number of mismatched bidders across true groups and simulated samples. HAD(λ) is the hazard rate of average discrepancy. For example, HAD(.10) = 0.002 means that in 499 simulated samples (out of a total of 500) the average number of mismatched bidders is less than 10 percent of the total number of bidders. Note:K 0 , EAD and HAD(λ) are defined as in Table 2. D µ is the difference between group means µ 1 and µ 2 . Conditional on the number of markets (L) and the number of bidders in population (n), the classification task is harder when the difference between group means D µ is smaller. where T 1 = (N 1 1 , ..., N 1 K 1 ) and T 2 = (N 2 1 , ..., N 2 K 2 ) are ordered partitions of N and denotes set-difference. We evaluate our classification method using two criterion: (1) Expected Average Discrepancy (EAD) defined as E(δ(T,TK)) and (2) the hazard rate of EAD, i.e., HAD(λ) ≡ P {δ(T,TK) > λn} for 0 < λ < 1. Table 2 reports estimates when there is no unobserved heterogeneity among bidders (K 0 = 1). In this case, our procedure detects the absence of unobserved heterogeneity effectively. For a given n, there is a moderate increase in the accuracy of classification as L increases, both in terms of EAD and HAD(λ). Table 3 reports results for K 0 = 2 and K 0 = 4. In both cases, the estimates for K 0 are mostly correct. Estimation accuracy increases with the difference between group means. For a given number of groups, the performance in terms of EAD and HAD are both better with greater group differences and larger sample sizes. Misclassification errors tend to arise less often when the number of true groups is smaller, with the exception of D µ = 0.2 and L = 100. Intuitively, this is because when the set of bidders is partitioned into fewer groups, we can use pairwise inequalities for classification. Two-Step Estimation in a Structural Model In this section we use a simple structural model of procurement auctions to investigate the impact of classification errors on subsequent estimation of structural parameters. A set of N providers (bidders) is partitioned into K 0 groups, each with a distinct distribution of private costs. Let N k denote the set of providers in N with type k ∈ {1, 2, ..., K 0 }, and let \|N k \| denote its cardinality. The cost for a provider i with type Auction participants are determined in two steps. First, two out of K 0 groups, τ l,1 and τ l,2 , are chosen at random. Next, n 1 and n 2 providers are randomly drawn from the corresponding groups N τ l,1 and N τ l,2 and their costs were constructed as above. Here n 1 and n 2 denote the numbers of actual participants (those who submitted bids in the auction). Then participants bid based on their realized costs. The participant with the lowest bid wins. The identity of each participant and its bid are both reported in data. We consider two specifications: (S 1 ) with K 0 = 4, \|N k \| = 4 for all k, and (S 2 ) with K 0 = 4, \|N k \| = 10 for all k. For both specifications, we set µ = (2, 2.4, 2.8, 3.2) and σ = 0.5. We run the following four experiments with different specification and sample sizes: (A) S 1 , L = 200; (B) S 2 , L = 200; (C) S 1 , L = 400; and (D) S 2 , L = 400. We set n 1 = 1 and n 2 = 1 so that each auction has 2 actual participants chosen from K 0 different types. Structural parameters K 0 , τ (·), θ = {(µ k ) K 0 k=1 , σ} are estimated via two steps. First, the group structure (τ (·) andK) are estimated using our classification algorithm. Next, we apply GMM to estimate the remaining structural parameters using the following moments for k = 1, ..., K 0 : (1) within-group means of bids: (2) within-group second moment of bids: where µ B,k (θ; I) and σ B,k (θ; I) denote the mean and standard deviation of equilibrium bid distribution for bidders from group k, θ the vector of parameters and I the profile of participant types. Standard errors are computed from the analytic expression for the covariance matrix in asymptotic distribution. To compute µ B,k (θ 0 ; I) and σ B,k (θ 0 ; I) for a given θ 0 and profile of participant types I = (τ l,1 , τ l,2 , n 1 , n 2 ) we simulate the equilibrium bidding functions. 15 The bidding functions 14 We set the upper and lower bounds of costs to c = 1 True parameters are chosen so that c is strictly positive. 15 Specifically, we start from the analytical bidding function when all participants belong to the same group, and use a modified version of the numerical method in Marshall, Meurer, Richard, and Stromquist (1994) Note: Specification A uses K 0 = 4, n k = 4, and L = 200 and Specification B uses K 0 = 4, n k = 10, and L = 200. Here n k is the number of bidders in group k, and L the number of markets. The rejection probabilities are from t-tests for the individual parameters. The nominal rejection probability is set to 0.05. are then combined with the cost distributions implied by a vector of trial parameters, θ 0 , to obtain the distribution of bids: . Here F C,k (.\| θ 0 ) denotes the distribution of project's cost for a bidder belonging to group k which is corresponds to a parameter vector θ 0 and β k (c), β −1 k (b) are the bid and the inverse bid functions used by such bidder. We then compute the mean and the standard deviation of this bid distribution. Tables 4 and 5 report the bias and mean squared errors (MSEs) of two estimators for (µ k ) K 0 k=1 and σ. The first is an "infeasible" estimator that uses the knowledge of the true group structure. The second is the two-step estimator we propose, which requires bidder classification in the first step. These two tables also report the rejection probabilities from t-tests of individual parameters. Table 4 reports results for a smaller sample with L = 200. The rejection probabilities are close to the nominal rejection rate 0.05, except for parameters µ 1 and σ. The MSE and bias for all parameters are reasonably small. Table 4 also shows that the rejection probabilities for the infeasible estimator using the true groups and those for the actual estimator using the estimated groups are very similar. There is some minor difference to solve for bidding strategies in the presence of multiple groups. We impose a sample version of c andc by replacing σ with its sample analog. Note: Specification C uses K 0 = 4, n k = 4, and L = 400, and Specification D, K 0 = 4, n k = 10, and L = 400. Here n k is the number of bidders in group k and L the number of markets. The nominal rejection probability is set to 0.05. between these two estimators in the MSE of some group means. The discrepancy seems more pronounced when the size of each group is increased from \|N k \| = 4 to \|N k \| = 10. Table 5 reports the same results for larger samples with L = 400. The performance of the estimators improves slightly relative to Table 4. Again, the rejection probabilities for the two estimators are similar. There is also evidence that with a larger number of the markets, our classification method performs better given the same number of withingroup bidders. Overall, Table 4 and 5 provide simulation evidence that the classification errors in the first-step do not have any major impact on the finite sample performance of the two-step estimators for structural parameters. Empirical Application: California Market for Highway Procurement We apply our methodology to analyze procurement auctions conducted by the California Department of Transportation (CalTrans) to allocate projects for highway repair work. Our goal is to demonstrate the performance of our method in the empirical setting, and to highlight the consequences of ignoring agent unobserved heterogeneity in estimation. Model. We follow the literature in modeling the auction market so that each project attracts a set of potential bidders who decide whether to participate in the auction and, if deciding to participate, choose a bid to submit. Our main innovation is to allow for contractors participating in this market to differ in a way that is not observed by the researcher. Specifically, each contractor is characterized by a contractor-specific cost factor (invariant across projects) q i which takes discrete values in {q 0 ,q 1 , ...,q K 0 }. This unobserved cost factor captures the difference in cost efficiencies across firms generated perhaps by the differences in managerial ability or other factors associated with the firm organization. As in our basic set up this cost factor induces partitioning of the population of firms participating in this market into the groups: Following the convention in the empirical auction literature, we assume that each project l auctioned in this market is summarized by a set of observable characteristics X l and an unobservable factor U l . The latter is distributed according to normal distribution with mean zero and standard deviation σ U . The set of firms which are potentially interested in project l (potential bidders), denoted here by S l , is exogenously drawn from N . A contractor i that is a potential bidder for project l is characterized by private entry costs, E i,l , and the private cost of completing the project, C i,l . We assume that private costs vary independently across bidders and auctions. The entry costs additionally are independent of U l , and are distributed according to the exponential distribution with a rate parameter λ i,l . The costs of completing the work are drawn from a log normal distribution with mean µ i,l and standard deviation σ C . The mean of the cost distribution depends on project characteristics including the distance between the project and the bidder's locations, D i,l , as well as an unobserved cost factor q i . 16,17 Reflecting these features, we set µ i,l = X l α 1 + D i,l α 2 + K 0 k=1q k 1{τ (i) = k} + U l , and λ i,l = X l γ 1 + K k=1q k 1{τ (i) = k}. 16 The groups reflect differences in the contractors' cost efficiencies related to the project work. While entry costs may also vary across groups, there is no reason for the group differences in project costs to coincide with the group differences in entry costs. For this reason, we explicitly distinguish between the parameters capturing the former (q k ) and the latter (q k ) effects. 17 Following the literature, we distinguish between the bidders who regularly participate in the procurement market (regular bidders) and those who only appear in a very small number of auctions (fringe bidders). We assume that all fringe bidders are associated with the same fixed level of the unobserved cost factorq 0 . A potential bidder decides to participate in the auction for project l if his ex-ante expected profit conditional on participation exceeds entry costs. 18 A bidder who decides to participate observes realization of his costs and the identities of other contractors who decided to participate. He chooses a bid to maximize his interium profit which reflects probability of winning the project conditional on his costs draw and the set of competitors. We assume that the observed outcomes reflect a type-symmetric pure-strategy Bayesian Nash equilibrium (psBNE). 19 Estimation Details. The estimation methodology consists of two steps. In the first step we use the pairwise comparison indexes to recover the unobserved group structure. In the second step the parameters of the model are estimated through a GMM procedure while imposing the group structure recovered in the first step. We assume bids are rationalized by a single equilibrium. In the first step we use the pairwise comparison indexes derived in Appendix B.1 in the supplemental note to recover the unobserved group structure. 20 Specifically, in accordance with the notation used in the paper, we define δ ij ≡ max {r ij (b), 0} db and 21 We obtain empirical counterparts of these indexes by replacing G ij (b\|d) with its sample analogĜ ij (b\|d). We implement classification using the bootstrap testing procedure described previously. In the second stage, we consider the following moments: (a) the first and the second moment of bid distribution for a given level of d and for a given group of bidders; (b) the covariance between bids and the observable project characteristics; (c) the covariance between any two bids submitted in the same auction; (d) the expected number of participants in any given auction for every (d, q)-group; (e) the covariance between the number 18 The expected profit reflects his expectation over the participation decisions of other potential bidders, the expectation over his costs of completing the project, and reflects expected probability of winning the project which depends on the costs draws of his competitors. 19 In such an equilibrium, participants who are ex ante identical in an auction l (i.e. i, j ∈ S l such that q i = q j , and D i,l = D j,l ) adopt the same strategies. 20 The pairwise comparison indexes are derived using Corollary 3 of Lebrun (1999) for all b in the common support of B i,l and B j,l whenever τ (i) ≤ τ (j). The inequality holds strictly at least over some interval with positive Lebesgue measure and holds unconditionally when aggregated over bidder identities and auction characteristics. 21 We recover group structure on the basis of the indexes which aggregate over the values of the distance d. As a robustness check we also compute groupings on the basis of subsets of distances. We find that the results of classification are very similar across these approaches. of participants and the observable project characteristics. We search for the set of parameters which minimizes the distance between the empirical and theoretical counterparts of these moments subject to participation constraints. 22 Estimation Results. We implement the analysis using the data for California Highway Procurement projects auctioned between 2002 and 2012. The projects in our sample are worth $523,000 and last for around three months on average; 38% of these projects are partially supported through federal funds. There are 25 firms that participate regularly in this market. The other firms are referred to as "fringes". An average auction attracts six regular potential bidders and eight fringe bidders. Since only a fraction of potential bidders submits bids, an entry decision plays an important role in this market. Finally, the distance to the company location varies quite a bit and is around 28 miles on average for regular potential bidders. In the first step, we obtain through our classification method the grouping of the bidders into eight groups that consist of 2, 3, 8, 3, 2, 3, 2 and 2 bidders respectively. The parameter estimates obtained in the second stage of our estimation procedure and their standard errors are summarized in Table 6. We normalize bids by the engineer's estimate in the estimation. Therefore all the parameters measure the effects relative to the project size. The first two columns present the estimates which are obtained when the unobserved group structure is taken into account in the estimation. The results indicate significant differences in bidders' costs across the groups. Specifically, fringe bidders (the reference group) tend to have the highest costs whereas the difference in costs between the group of fringe bidders and the groups of regular bidders is comparable in impact to the shortening of the distance to the project site by 42.5 (i.e., by 0.051/0.0012), 10.1, 26.67, 48.33, 11.67, 6.67, 7.5, and 41.67 miles respectively. 23 The distance increases project costs (additional 8.33 miles result in costs which are 1% higher on average). 24 The entry costs of regular bidders are significantly lower than entry costs of fringe bidders. However, they appear to be quite similar across the groups of regular bidders. The last two columns of Table 6 show the parameter estimates under the specification when the unobserved group structure of the regular bidders is ignored in the estimation. The parameter estimates are obtained by the GMM estimation procedure using the same 22 We do not explicitly solve for participation strategies. Instead, we discretize the support of auction characteristics (X l , U l ) and treat the probabilities of participation for bidders of various (q, d)−types corresponding to these grid values as auxiliary parameters. We follow the spirit of Dube, Fox, and Su (2012) by maximizing a moment-based objective function subject to the constraints that the optimality of the participation strategies is satisfied on the grid of the project characteristics' values. 23 The estimates of the group-specific fixed effects are omitted for brevity. The full table that contains these estimates is found in the supplemental note to the paper. 24 Recall that the coefficients reflect the impact on costs in terms of the fraction of the engineer's estimate. The distance resulting in 0.01 increase of average costs can thus be computed as 0.01/0.012. Note: In the results above the distance is measured in miles. The fringe bidders are the reference group. The first two columns correspond to the specification which allows for the unobserved bidder heterogeneity; the last two columns correspond to the specification without unobserved bidder heterogeneity. The results are based on the data for 1,054 medium-sized projects that involve paving and bridge work. Standard errors are computed using bootstrap. set of moments by imposing that only two groups of sellers are present in the data: fringe and regular bidders. Under this specification, the cost reduction due to the federal aid is estimated to be much higher (7.8% rather than 4.3%), the impact of the distance is estimated to be lower (the distance to the project has to be 11.67 miles higher in order to increase the average cost by 1%). Additionally, the entry costs are estimated to be lower relative to the baseline specification. Our results thus confirm that regular participants in the highway procurement market are characterized by important unobserved cost differences that persist in the data. Conclusion This paper makes a number of contributions to the literature. First, for models with strategic interdependence between multiple agents, we develop a method to classify these agents based on their discrete unobserved individual heterogeneity, using pairwise inequalities implied by an economic model. Second, we show such pairwise inequalities arise in a number of game-theoretical settings where identification of model primitives is challenging. Third, we propose a computationally feasible method which consistently estimates the group structure defined by unobserved heterogeneity. We apply this method to California highway procurement data to show that unobserved bidder heterogeneity plays an important role in this procurement market. The classification method proposed in this paper is especially useful in settings where the analysis of unobserved individual heterogeneity is complicated by the presence of strategic interdependence in the model. We offer new insights into the identification and estimation of such models. Specifically, classification could be used as a first step in the structural studies of many environments where analyses would otherwise be infeasible due to the identification or computational challenges. This note is a supplemental note to Krasnokutskaya, Song, and Tang (2019). It consists of five parts. Appendix B gives details about how to derive pairwise comparison index in examples of first-price and English auctions where bidders have asymmetric private values, or collusive behavior. Appendix C gives mathematical proofs of the results in the paper, and provide lower level conditions. Appendix D discusses a bootstrap method to construct a confidence set for the group structure. Appendix E presents further simulation results regarding the performance of our classification algorithm proposed in Krasnokutskaya, Song, and Tang (2019), and Appendix F reports summary statistics for the data used in the empirical application in the paper and some further results. B.1. First-Price Auctions with Asymmetric Bidders Let the population of bidders, N , be partitioned into K 0 groups, each of which is characterized by a distinct distribution of private values F k (·). To fix ideas, assume that F k (·) has the same shape of distribution, but differs only in their location (means)q 1 < q 2 < ... <q K 0 , withq k being the mean of F k . (Our method applies in a more general setting when F k 's are stochastically ordered.) In this case, for a bidder i with τ (i) = k the mean of the private value distribution is given by q i =q k . Let S l denote the set of participants in auction l and B l = {B i,l } a vector of bids. In a type-symmetric equilibrium B i,l = β k (V i,l ) if τ (i) = k, with β k (·) being a bidding strategy and V i,l the private value for i in auction l. Define G ij (b) = P {B i,l ≤ b\| i, j ∈ S l }, where S l denotes the set of bidders in auction l. We assume bids are rationalized by a single equilibrium. Corollary 3 of Lebrun (1999) showed that G ij (b) ≥ G ji (b) for all b in the common support of B i,l and B j,l whenever τ (i) ≤ τ (j). The inequality holds strictly at least over some interval with positive Lebesgue measure. This inequality holds unconditionally when aggregated over bidder identities and auction characteristics. 25 Thus pairwise comparison indexes can be constructed as follows: Likewise, define δ 0 ij by replacing the integrand in δ ij by the absolute value of G ij (b) − G ji (b). These indexes do not condition on the specific identities of bidders participating in each auction l. Thus it allows us to utilize observations from a large number of auctions when constructing a comparison index for any generic pair i and j. For the rest of this subsection, we derive the pairwise comparison inequalities in a general model of asymmetric first-price auctions where independent private values are drawn from distributions that are stochastically ordered. That is, F k first-order stochastically dominates F k whenever k > k. Also assume that the ordering of the distributions is strict (F 1 (v) > F 2 (v) > ... > F K 0 (v)) at least for v within some non-degenerate interval on the support. Let N (k) denote the set of all agents in group k. For simplicity, suppose that a bidder from group k becomes active with a fixed probability that is exogenously given. Let A denote the set of entrants in a given auction and λ(A) denote the structure, or the profile, of entrants. That is, λ(A) is a K 0 -vector of integers (\|A (1) \|, ..., \|A (K 0 ) \|), with A (k) being the set of entrants from group k. An entrant i submits bid B i according to his private value v i , taking into account the competitive structure of an auction λ(A) which he observes at the time of bidding. Across auctions in the data, A and {v i } i∈A are independent draws from the same population distribution. Let G k (·; λ) be the distribution of B i when i ∈ N (k) . The private values are independent of λ A under exogenous entry. Part (i) of Corollary 3 in Lebrun (1999) showed that, given any realization of λ(A), the supremum of the support of bids is the same for all bidder types. That is, for any λ, β 1 (v\|λ) = β 2 (v\|λ) = ... = β K (v\|λ) ≡ η(λ) < ∞ for some η(λ) ∈ (v, v), where β k denotes the equilibrium bidding strategy for a bidder from group k. Furthermore, the corollary also showed that for any λ(A), F k (β −1 k (b\|λ(A))) ≤ F k (β −1 k (b\|λ(A))), for all b ∈ [v, η(λ(A))] and k < k , and the inequality holds strictly at least over some interval on [v, η(λ(A))]. Consider i ∈ N (k ) and j ∈ N (k) with k > k. It then follows that with the inequality holding strictly over some non-degenerate interval in the shared bid support. The inequality does not condition on the identities of the entrants other than i and j. Finally, note that by a symmetric argument, a similar inequality holds in first-price procurement auctions with P {B i ≥ b\|i, j ∈ A} ≤ P {B j ≥ b\|i, j ∈ A} (with inequality being strict over some non-degenerate interval in the shared bid support), whenever the private cost distribution for i is stochastically lower than that of j. B.2. English Auctions with Asymmetric Bidders Consider the setting in Section B.1, except that the auction format is English (ascending). The data report the identities of entrants in A and the transaction price W in each auction. In a dominant strategy equilibrium, the price in an auction equals the second highest private value among all entrants. With independent private values, we show below that for all w over the intersection of support, whenever τ (i) > τ (j). Furthermore, the inequality holds strictly for some w over a set of positive measure in common support. This implies The intuition behind (B.2) is as follows. Given any structure of entrants who compete with i or j (but not both), the distribution of the transaction price is stochastically higher when i is present but j is not than when j is present but i is not. Loosely speaking, when j is replaced by the stronger type i in the set of entrants, the overall profile of value distributions becomes "stochastically higher". Then the law of iterated expectations implies (B.2) and (B.3). To infer the group structure, define the following indexes: One can then use our procedure proposed in the main text to classify the bidders based on pairwise comparison. We now derive (B.2) formally. Let V i denote the private value for bidder i. Consider the case where i ∈ N (k ) and j ∈ N (k) with k > k. Let λ(A) denote the K 0 -vector of integers that summarizes the group structure of the set of entrants A. Let 1 k denote the unit vector with the k-th component being 1. Then define where the first term on the right-hand side equals F k (w) , and the second on the right-hand side is where ϕ(w; λ * ) denotes the probability that the maximum value in A\{j} is strictly greater than w while the second highest value in A\{j} is less than or equal to w conditional on the classification λ(A\{j}) = λ * . Therefore By the same argument, It is then straightforward to show that for any λ * , that F k F.S.D. F k implies H i,j (w; λ * ) ≤ H j,i (w; λ * ) over the union of the K 0 supports of {F l : 1 ≤ l ≤ K 0 }, and the inequality holds strictly at least for some w in an interval on the intersection of the K 0 supports of {F l : 1 ≤ l ≤ K 0 }. Under exogenous entry, we get after integrating out λ * . The inequality holds strictly for some w over common support. One may wonder whether we can recover the classification of bidders in the English auction example through a "global" approach when the identity of the winner is reported in the data. That is, by comparing the distribution of transaction prices when i is the winner versus that when j is the winner, as opposed to the pairwise comparison approach proposed above. Let us explain why this is not feasible. For any i ∈ N (k ) and j ∈ N (k) and F k F.S.D. F k , let A\{i, j} denote the set of entrants out of N \{i, j} and let M (A\{i, j}) ≡ max{V s : s ∈ A\{i, j}}. Let φ(w; λ * ) denote the distribution of M (A\{i, j}) conditional on λ(A\{i, j}) = λ * . Let D denote the identity of the winner in the auction; and S k denote the survival function for the private value of a type-k bidder. Then, where p j is shorthand for j's entry probability. Also note that, by construction, once conditioned on the realized set of entrants from A\{i, j}, we have Likewise P {W ≤ w, D = j\|j ∈ A} can be written by swapping the roles of i and j and swapping the roles of k and k respectively. Then it can be shown that To see why the inequality in (B.4) holds, note for any λ * , where the difference is written as The first square bracket in the display above is positive because Furthermore, the second square bracket in the display is also positive because "F k F.S.D. F k " implies S k (w) ≥ S k (w) and F k (w) ≥ F k (w) for all w and these inequalities hold strictly for some set of w with positive measure. Integrating out λ * on both sides of the inequality yields the first inequality in (B.4). Similarly, the difference between P {W ≤ w, D = i\|i ∈ A, j ∈ A, λ(A\{i, j}) = λ * } and which must be positive because the two terms in the square brackets are positive. Now we write where p j ≡ P (j ∈ A). A similar expression exists for P {W ≤ w, D = j\|j ∈ A} by swapping the roles of i and j in (B.5). Therefore the difference between P {W ≤ w, D = i\|i ∈ A} and P {W ≤ w, D = j\|j ∈ A} is also indeterminate in the absence of knowledge about p i , p j . B.3. Bidding Cartel in First-Price Procurement Auctions Our method can be used to detect the identities of cartel members in a model of firstprice procurement auctions in which a bidding cartel competes with competitive noncolluding bidders (Pesendorfer (2000)). Let the population of bidders/firms N be partitioned into a set of colluding firms N (c) and non-colluding firms N (nc) . In each auction, the set of potential bidders (who are interested in bidding for the contract) A is partitioned into A (c) and A (nc) . The cardinality of A (c) is common knowledge among the bidders. The potential bidders in A (c) collude by refraining from participation except for one bidder i * who is chosen among them to submit a bid. 27 In an efficient truth-revealing mechanism considered in Pesendorfer (2000), the cartel member that has the lowest cost is selected to be the sole bidder from the cartel. That is, i * (A (c) ) = arg min j∈A (c) C j where C j is the private cost of bidder j. Thus, the set of final entrants who are observed to submit bids in the data is A * ≡ {i * (A (c) )} ∪ A (nc) . (The set of colluding potential bidders is not reported in the data available to the researcher.) We maintain that across the auctions in the data bidders' private costs are independent draws from the same distribution. Bidders are ex ante symmetric in that each bidder's private cost is drawn independently from the same distribution. Entrants know that a representative of the cartel is participating in bidding, and all follow Bayesian Nash equilibrium bidding strategies. We are interested in detecting the identities of the set of colluding firms in N (c) from the reported bidding and participation decisions. Let N (c) ⊂ N denote the set of bidders such that no two bidders in N (c) are ever observed to compete with each other in the bidding stage. By construction, N (c) ⊆ N (c) so the latter should be interpreted as a set of suspects for collusion. However, the set N (c) could also contain innocent noncolluding bidders who are never observed to compete with each other in the data because of random entry in finite sample. Our goal is to use bidding data to separate N (c) from Pesendorfer (2000) (Remark 3) shows that in any given auction with participants A (c) ∪ A (nc) , the distribution of bids from a non-colluding bidder j first-order stochastically dominates the distribution of the bids from the sole bidder representing the cartel i * . 28 Specifically, for any such i * and j, for all b on the common support of the two distributions. 29 The intuition, as is noted in Pesendorfer (2000), is that the sole bidder representing a cartel has a higher hazard rate than a non-colluding bidder. That is, relative to a competitive bidder, the cartel representative has a higher probability of having a low cost conditional on the costs being above any fixed threshold. Besides, ex ante symmetry among bidders implies that We can then construct pairwise comparison indexes δ ij , δ 0 ij by replacing G ij and G ji in (B.1) with the left-and right-hand side of (B.6). C.1. Proof of Identification Results Proof of Theorem 3.1: Sufficiency is obvious. We focus on necessity. First consider the two facts: Fact 1: If G τ does not contain a monotone path of length K 0 − 1, τ is not identified. Fact 2: A vertex i is identified if and only if there is a monotone path P containing i such that its end vertices i H and i L are identified and where (P ) denotes the length of P . By Fact 1, the necessity of G τ containing a monotone path of length K 0 − 1 follows, and Fact 2 completes the proof of the necessity part of the theorem. Now let us prove Fact 1. Suppose that G τ does not contain a monotone path of length K 0 − 1. Let N max be the set of vertices such that for each vertex i in N max , all his G τneighbors have lower type than the vertex i. Then there is no edge in G τ which joins any two vertices from the set N max . Choose a vertex i * from N max which is an end vertex of a longest monotone path, say, with length K − 1 < K 0 − 1. This identifies a lower bound for K 0 but there is no upper bound for K 0 that we can obtain from G τ . Take any τ such that τ (i * ) > τ (i * ) and τ (i) = τ (i) for all i ∈ N \ {i * }. Then τ is compatible with G τ and the given comparison indexes, proving that τ is not identified from G. Let us prove Fact 2. Sufficiency is trivial. Let us focus on necessity. First, suppose to the contrary that there is no monotone path with identified end vertices which contains i. Then obviously i is not identified. Therefore, if i is identified, there exists a monotone path with identified end vertices which contains i. So it suffices to show that if i is identified, it is necessary that such a monotone path has to have length equal to τ (i H ) − τ (i L ). Suppose to the contrary that every monotone path P that contains i and has identified end vertices i H and i L also satisfies τ (i H ) − τ (i L ) > (P ). Then we will show that i is not identified. First, assume that there exists a monotone path which contains i but not as one of its end vertices. Let i * H be a lowest type vertex among all the identified vertices each of which is on a monotone path that contains i and is of higher type than i. Also, let i * L be a highest type vertex among all the identified vertices each of which is on a monotone path that contains i and is of lower type than i. Let P be a monotone path between i * H and i * L that passes through i. Then by construction, the type difference τ (i * H ) − τ (i * L ) between the two end vertices is smallest among all the monotone paths that go through i. By the condition, we have τ (i * H ) − τ (i * L ) > 2. Therefore, we have multiple different ways to assign τ (i * H ) − 1, τ (i * H ) − 2, ..., τ (i * L ) + 1 to the vertex i on the path P . Hence i is not identified. Second, assume that all the monotone paths that contain i have i as one of their end vertices. Then either all neighbors of i are of higher type than i or all neighbors of i are of lower type than i. Suppose that we are in the former case. (The latter case can be dealt with similarly.) Let i * H be a lowest type vertex among all the vertices each of which is on a monotone path that contains i and is of higher type than i. Then by the condition, τ (i * H ) − τ (i) > 1. Thus we have multiple different ways to assign τ (i * H ) − 1, τ (i * H ) − 2, ..., τ (i) + 1, τ (i) to the vertex i. Hence i is not identified. C.2. Lower Level Conditions for Assumption 4.1 Lower level conditions for Assumption 4.1 can be found from the literature of testing for moment inequality restrictions. For the sake of concreteness, we focus on the situation where r ij (x) arises from difference between two nonparametric regression functions and the testing procedure is done by the method proposed in Lee, Song, and Whang (2018). Suppose that we have , i ∈ N , and = 1, ..., L, is the sample unit index. Let us define a kernel estimator of g i (x) as follows: where K h (x) = K(x/h)/h, and K(·) is a multivariate kernel and h is a bandwidth. We let Then the test statistics we use are defined as As for the bootstrap test statistics, we first obtain bootstrap samples with replacement. Using the bootstrap sample, we construct . Then the bootstrap test statistic we use is defined as 30 Let the CDF of the bootstrap distribution of T s * ij be denoted by F s ij . Then we set the pairwise p-value to bep s ij = 1 − F s ij (T s ij ). for construction of the pairwise p-values and for the implementation of the classification algorithm of this paper.) Then we can rewritê ij is the CDF of the bootstrap distribution ofT s * ij . Let · ∞ denote the sup norm, i.e., f ∞ = sup x \|f (x)\| for any real function f . Let us consider the following set of assumptions. Assumption C.1. (i) For each s ∈ {+, 0, −} and each i, j ∈ N , there exist sequences of constants a s ij,L and σ s ij,L such that the conditions in (C.4) hold, and The lower level conditions for Condition (i) can be found in Lee, Song, and Whang (2018). Condition (ii) follows if the kernel regression estimatorsĝ i (x) are uniformly consistent. (See, e.g., Hansen (2008).) Condition (iii) is a standard bandwidth condition in the literature of kernel estimators. Then, we obtain the following lemma. Lemma C.1. Suppose that Assumption C.1 holds, and that the sequence r L → ∞ is such that r L / log L → 0 as L → ∞. Then Assumption 4.1 holds. Proof: Condition (i) of Assumption 4.1 follows from Condition (i) of Assumption C.1 with W s ij being a standard normal random variable. As for Conditions (ii)-(iv) of Assumption 4.1, we focus on only (ii), because the proof for the other two statements is similar. Observe that when τ (i) > τ (j), so that by Assumption C.1 (ii). Hence Condition (ii) of Assumption 4.1(ii) follows with and λ L = √ L. As for Condition (vi) of Assumption 4.1, note that F s ij,∞ = Φ, the standard normal CDF. Hence there exists C > 0 such that for all t > C, Therefore, for any constants c 1 , c 2 > 0, taking λ L = √ L and ρ L = L −α , (from some large L on) as L → ∞, by the condition that r L / log L → 0. C.3. Proof of Consistency of Classification We provide a proof of consistency of the classification allowing both n and L to diverge to infinity jointly. This requires a more elaborated set of assumptions than Assumption 4.1. It is not hard to use the same proof to prove Theorem 4.1. Assumption C.2. There exist sequences ρ L , κ L → 0 and λ L → ∞ and constants c s ij > 0, s ∈ {+, 0, −} such that along each sequence of probabilities P L ∈ P 0,ε , the following holds for all s ∈ {+, 0, −}, s ∈ {+, −}, as n, L → ∞. ( Assumption C.3. Suppose that F s ij and F s ij,∞ , s, s ∈ {+, 0, −}, are CDFs and ρ L , κ L , and λ L are sequences in Assumption C.2. Then the following holds. ( Assumptions C.2 and C.3 are variants of Assumption 4.1 (except for Assumption C.3(iii)) which require the rate of convergences explicitly. Assumption C.3(iii) is new here because we now allow both n and L to increase to infinity. This assumption says that for the consistency of the estimated group structure, we need to have n increase sufficiently faster than L. This condition is trivially satisfied when n is fixed and L increases to infinity. Note that the conditions are high level conditions suited to the generic set-up here. With a more detailed specification of the pairwise comparison indexes δ ij and δ 0 ij , test statistics and p-values, one may improve the conditions. Using high level conditions also enables us to focus only on more novel aspects of the mathematical proofs. While existence of such sequences as κ L , ε L , and ρ L is fairly expected, obtaining their precise forms using lower level conditions require substantial yet tedious arguments depending on the way the test statistics and the p-values are constructed. Essentially what one needs to obtain for lower level conditions is the rate in the convergence of the test statistics to the limiting distribution both under the null hypothesis and under the local alternatives. For example, if one follows the approach of Lee, Song, and Whang (2018), the rate of convergence is ultimately delivered by a Berry-Esseen bound (for a sum of independent random variables) used to obtain the asymptotic normality of test statistics (under appropriate mean shifts). However, the final result comes, in combination with this, only with carefully derived rate results on asymptotically negligible terms that arise, among other things, due to the Poissonization technique used in the paper. While these additional developments are feasible, they do not add insights to the main idea of the paper. Hence we do not pursue details in this direction here. Let us first state the theorem. Theorem C.1. Suppose that Assumptions C.2 -C.3 hold, and that g(L) → ∞ and g(L)/r L → 0 as L → ∞. Then, for any ε > 0, along a sequence of probabilities P n,L from P 0,ε , P n,L {K = K 0 } → 1, as n, L → ∞, and the estimated group structureTK satisfies that as n, L → ∞, The proof of this theorem is long. We first prepare some auxiliary lemmas. Throughout this section, we assume that Assumptions C.2 and C.3 hold. First, for any subset N ⊂ N , we define N (i) = N \ {i}, and let We also defineN Following the convention, given a CDF G, we define its generalized inverse G −1 as G −1 (t) = inf{s ∈ R : G(s) ≥ t}, t ∈ (0, 1). (ii) For all i, j ∈ N such that τ (i) = τ (j), , and for all i, j ∈ N such that τ (i) = τ (j), Proof: (i) We will show the first and the second statements only. The third and the fourth statements can be proved similarly. Let us prove the first statement first. For all i ∈ N , Note that where the last O(ρ L ) term is due to Assumption C.3(i) and Markov's inequality. As for the leading probability on the right end side, we use the fact that for any CDF G and any t ∈ (0, 1) and x ∈ R, G −1 (t) ≥ x if and only if t ≥ G(x), (e.g. Lemma A.1.1. of Reiss (1989)), and bound by Assumption C.2(ii)(b) and because j ∈ N 1 (i). By Assumption C.3(i), the last term is equal to Thus we obtain the first statement. Let us consider the second statement. Suppose that τ (i) > τ (j). We let and let A L = A 1,L ∩ A 2,L . Note that , where the last O(ρ L ) term is due to Assumption C.2(ii)(a) and Assumption C.3(i). For the leading probability on the right hand side, note that The last indicator becomes zero from some large L on by Condition (ii) of Assumption C.3. Thus, we obtain the second statement. (ii) The proof of the first statement is the same as that of the second statement of (i), and the proof of the second statement is the same as that of the first statement of (i). Details are omitted. Lemma C.3. Suppose that N ⊂ N contains agents with heterogeneous types. Then, for each i ∈ N , Proof: We show only the first and the third statements. The remaining statements can be proved similarly. Define Note that where the last inequality follows by the first and the second statements of Lemma C.2(i). Thus we have as n, L → ∞. On the other hand, The second inequality follows because logp − ij ≤ 0 and the last equality follows by the third statement of Lemma C.2(i). Since N 1 (i) and N 2 (i) partition N (i), andN 1 (i) and N (i) \N 1 (i) also partition N (i), it follows that P {N 1 (i) =N 1 (i)} = 1 + O(nω n,L ), and P {N 2 (i) = N (i) \N 1 (i)} = 1 + O(nω n,L ), as n, L → ∞. The remaining statements can be shown similarly. (ii) Let N τ (N ) denote the set of τ -ordered partitions of N . When an ordered partition (N 1 , ..., N s ) is a τ -ordered partition, and τ partitions N into K groups (i.e., any two agents, say i, j, from two different groups from the K groups have τ (i) = τ (j)), we must have s ≤ K by the definition of τ -ordered partition. Hence some group in the ordered partition (N 1 , ..., N s ) can have agents with different types. Lemma C.5. Suppose that an estimated ordered partition (N 1 , ...,N s ) of N is asymptotically τ -compatible at rate u n,L . Then the Selection Step in the Selection-Split Algorithm applied to this ordered partition selects with a set, say,Nr, such that P {∃i, j ∈Nr, τ (i) = τ (j)} = 1 + O(n 2 ω n,L + u n,L ), as n, L → ∞. Lemma C.6. For any set N ⊂ N with agents of heterogeneous types, the ordered partition (N 1 ,N 2 ) of N obtained by the Split Algorithm is asymptotically τ -compatible at rate n 2 ω n,L . Lemma C.7. Suppose that for some s < K 0 , an estimated ordered partition (N 1 , ...,N s ) of N is asymptotically τ -compatible at rate u n,L . Then a new ordered partition (N 1 , ...,N s+1 ) of N obtained by applying the Selection-Split Algorithm to (N 1 , ...,N s ) is asymptotically τ -compatible at rate n 2 ω n,L + u n,L . For each K ≥ 1, letT K = (N 1 , ...,N K ) be the group structure obtained through the Selection-Split algorithm (until the number of the groups reach K) and let T = (N 1 , ..., N K 0 ) be the true ordered group structure. For the remainder of the proof, we assume that the conditions of Theorem C.1 holds. (ii) If K < K 0 , then for any M > 0, as n, L → ∞, Proof: (i) Let (N 1 , ...,N K 0 ) be the ordered partition obtained by the Selection-Split Algorithm. By Lemma C.8, the event that τ (i) = τ (j) for all i, j ∈N k has probability approaching one for all k = 1, ..., K 0 . Let (N 1 , ...,N K ) be the ordered partition obtained by the Selection-Split Algorithm with K ≥ K 0 . Since K ≥ K 0 , due to the sequential split nature of the algorithm, each of the resulting groups, say,N k , k = 1, ..., K, is a subset of a group, say,N r , obtained at step K = K 0 . Therefore, the event that τ (i) = τ (j) for all i, j ∈N k has probability approaching one for each k = 1, ..., K. By Assumption C.2(i), we havê as n, L → ∞. Thus (i) follows. Proof: Choose K such that K 0 < K and writê As for the leading term on the left hand side, we havê by Lemma C.9(i). Since g(L) → ∞, we find that whenever K > K 0 , we have P Q (K 0 ) <Q(K) → 1. And for all K < K 0 , we have by Lemma C.9(ii), for any M > 0, whereasV (K 0 ) = O P (1). Therefore, choose ε > 0 and take any M > 0 such that We take large M > K 0 − K and 0 < M ≤ (K − K 0 + M )g(L). We find that as L → ∞. By sending ε down to zero, we conclude that P {K = K 0 } → 1, as n, L → ∞. Proof of Theorem C.1: The desired result follows from Lemmas C.8 and C.10. Appendix D. Confidence Sets for the Group Structure The web appendix of Krasnokutskaya, Song, and Tang (2018) proposes a method to construct a confidence set for each group of agents having the same type. Here for the sake of readers' convenience, we reproduce the procedure here using the notation of this paper. Let us consider a set-up where we have K 0 groups and the set N of agents. LetK be the consistent estimator of K 0 as proposed in Krasnokutskaya, Song, and Tang (2019). As for confidence sets, we construct a confidence set for each group of agents who have the same type. First, we fix k = 1, ...,K and construct a confidence set for the k-th type group N k . In other words, we construct a random setĈ k ⊂ N such that For this, we need to devise a way to approximate the finite sample probabilities like P {N k ⊂Ĉ k }. Since we do not know the cross-sectional dependence structure among the agents, we use a bootstrap procedure that preserves the dependence structure from the original sample. The remaining issue is to determine the space in which the random set C k ⊂ N can take values in. It is computationally infeasible to consider all possible such sets. Instead, we proceed as follows. First we estimateN k as prescribed in the paper and also obtainδ 0 ij , the test statistic defined in the main text. Given the estimateN k , we construct a sequence of sets as follows: Step 1: Find i 1 ∈ N \N k that minimizes min j∈N kδ 0 i 1 j , and constructĈ k (1) =N k ∪ {i 1 }. Repeat Step m up to n = \|N \|. Now, for each bootstrap iteration s = 1, ..., B, we construct the setsN * k,s and {Ĉ * k,s (m)} following the steps described above but using the bootstrap sample. (Note that this bootstrap sample should be drawn independently of the bootstrap sample used to construct bootstrap p-valuesp ij in the classification.) Then, we compute the following: Note that the sequence of setsĈ * k,s (m) increases in m. Hence the numberπ k (m) should also increase in m. An (1 − α)%-level confidence set is given byĈ * k (m) with 1 ≤ m ≤ n such thatπ k (m − 1) < 1 − α ≤π k (m). Note that such m always exists, becauseĈ * k,s (n) = N . Appendix E. Further Simulation Results Tables E.1 and E.2 summarize the distribution of estimation errors in our group classification algorithm from 500 simulated data sets, when the number of groups is K 0 = 2 and assumed known to the econometrician. The column D µ shows the difference between the group means chosen in the simulation. When K 0 = 2, the results show that the estimation error, as measured by the expected average discrepancy (EAD), decreases with the distance between group means. Such a reduction in EAD is most substantial when the number of agents is larger (n = 40) and the size of the data is small (L = 100). Given group difference, EAD decreases as sample size increases moderately from L = 100 to 400. This pattern is most obvious when D µ = 0.2. The other measure of estimation errors, HAD(p), also shows encouraging results. HAD(p) is zero for most of the cells in both panels (a) and (b), which shows that the empirical distribution of proportion of mis-classified bidders is reasonably skewed to the right. Besides, the reduction in HAD(p) as the sample size increases is most pronounced with closer group means, regardless of the number of bidders in the population. When K 0 = 4, the results demonstrate very similar patterns. Most remarkably, both measures of mis-classification errors only increase very marginally relative to the case with K 0 = 2. Tables E.3 and E.4 report results from the full, feasible classification procedure when the number of groups is estimated through the penalization scheme proposed in the text. For most of the specifications used in these two tables, the estimates for the number of groupsK 0 are tightly clustered around the correct K 0 . Compared with the results for infeasible classification under known K 0 , EAD and HAD(p) increase in most cases. Nonetheless such an increase is quite moderate, suggesting that our feasible classification algorithm performs reasonably well relative to its infeasible counterpart. In Tables E.3 -E.4, we report the analysis of computation time for the classification procedure. In Table 5.3, we give a decomposition of the time that it took for the classification procedure. The table clearly shows that the major computation time spent is when we construct bootstrap p-values. Once the p-values are constructed, the classification algorithm itself runs fairly fast. In Table E.4 , the computation time is shown to vary depending on the number of the agents (n), the number of the true groups (K 0 ), and the number of the markets (L). The results show that the most computation time increase arises when the number of the bidders increases rather than when the number of the markets or the number of the groups increases. Our simulation studies are based on our MatLab code. The program was executed using a computer with the following specifications: Intel(R) Xeon (R) CPU X5690 @3.47 GHz 3.46 GHz. Note: This table summarizes the distribution of estimation errors in our classification algorithm from 500 Monte Carlo replications when K 0 = 4 and known. Here n represents the number of the individual agents, L the number of the observed games in the data, D µ the distance between population means, EAD the expected average discrepancy, and HAD(p) the hazard rate of average discrepancies at p. Note: This table summarizes the distribution of estimation errors in our classification algorithm from 500 Monte Carlo replications when K 0 = 4 and known. Here n represents the number of the individual agents, L the number of the observed games in the data, D µ the distance between population means, EAD the expected average discrepancy, and HAD(p) the hazard rate of average discrepancies at p. Step Description L=100 L=200 L=400 L=100 L=200 L=400 L=200 L=200 Note: The table shows the change in the computation time as one changes the number of the groups (K), the number of the markets and the number of the agents (i.e., bidders) ((n)). The major increase in the computation time arises when the number of the bidders increases rather than when the number of the markets or the groups increases. Appendix F. Additional Materials for the Empirical Application Table F.1 reports summary statistics for this set of projects. The table indicates that the projects are worth $523,000 and last for around three months on average; 38% of these projects are partially supported through federal funds. There are 25 firms that participate regularly in this market. All other firms in the data are treated as fringe bidders. An average auction attracts six regular potential bidders and eight fringe bidders. Since only a fraction of potential bidders submits bids, an entry decision plays an important role in this market. Finally, the distance to the company location varies quite a bit and is around 28 miles on average for regular potential bidders. Note: This table reports summary statistics for the set of medium size bridge work and paving projects auctioned in the California highway procurement market between years of 2002 and 2012. It consists of 1,054 projects. The distance variable is measured in miles. It reflects the driving time between the project site and the nearest company plant. The "Federal Aid" variable is equal to one if the project receives federal aid and zero otherwise.	2019-08-04T04:58:24.000Z	2019-08-04T00:00:00.000	{ "year": 2021, "sha1": "01224ce9c80b681e3f28f98d1810f24182303bf2", "oa_license": null, "oa_url": "http://arxiv.org/pdf/1908.01272", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "01224ce9c80b681e3f28f98d1810f24182303bf2", "s2fieldsofstudy": [ "Economics" ], "extfieldsofstudy": [ "Computer Science", "Economics", "Mathematics" ] }
43775260	pes2o/s2orc	v3-fos-license	Reduced cell attachment and phosphorylation of focal adhesion kinase associated with expression of a mutant insulin receptor. Insulin signaling results in rapid changes to the cell cytoskeleton, and it has recently been shown that insulin stimulates the dephosphorylation of the cytoskeletal-associated tyrosine kinase, focal adhesion kinase (pp125FAK). We report here that mutation of two tryptic cleavage sites (Lys164 and Lys582 → Asn; 2N) in the insulin receptor α-subunit results in a cell-line (CHO.2N-10) with altered morphology associated with an increase in cell size, a decrease in cell adhesiveness, and a decrease in pp125FAK tyrosine phosphorylation in the absence of insulin (45.2 ± 9.7% compared to nontransfected Chinese hamster ovary (CHO) cells). In contrast to pp125FAK, paxillin phosphorylation was similar in all cell lines despite lower levels (61.0 ± 10.4% compared to CHO cells) of paxillin protein in CHO.2N-10 cells. We observed comparable protein levels of pp125FAK and the structural focal adhesion protein, vinculin, in all cell lines. Despite underphosphorylation of pp125FAK in the basal state, insulin stimulation of CHO.2N-10 cells still resulted in dephosphorylation of pp125FAK. CHO.2N-10 and CHO.T (overexpress wild-type insulin receptor) cells have similar insulin binding characteristics, insulin-stimulated autokinase and peptide phosphorylation, and insulin-stimulated pp185/IRS-1 phosphorylation. Our results suggest that the insulin receptor may play an important role in cell-matrix interactions, such as modulating cell adhesion and inducing cell architecture changes. Interactions between growth factors and integrin signaling pathways are probably involved in the regulation of cell proliferation, shape, adhesion and migration (1,2). Integrins, the receptors for extracellular matrix proteins, provide both a physical link to the cytoskeleton (often at sites of focal adhesion) and transduce signals from the extracellular matrix. A connection is observed between integrins and stress fibers of polymerized actin, which is necessary for maintenance of cell integrity and appears to terminate at focal adhesion sites. The assemblies of structural proteins (such as ␣-actinin, talin, vinculin, tensin, and paxillin) that co-localize with some integrins in focal adhesions are thought to play important roles in stabilizing cell adhesion and regulating cell shape, morphology, and mobility (1). These structural proteins may also serve as a framework for the association of signaling molecules that regulate signal transduction pathways leading to integrin-induced changes in cells (1). In addition, there are now many examples of growth factor-induced regulation of components involved in integrin signaling pathways. Growth factors may rapidly stimulate actin polymerization at the plasma membrane of many cell types to produce lamellipodia and edge-ruffles (e.g. PDGF 1 and insulin) (3)(4)(5) and at later times promote actin stress fiber formation (e.g. PDGF) (6). Conversely, components of growth factor signaling pathways (such as the mitogen-activated protein kinase cascade) are probably utilized by integrins as part of their signal transduction pathways (7,8). A key enzyme in cytoskeletal rearrangements/interactions/ architecture appears to be the phosphorylated, cytoplasmic tyrosine kinase, focal adhesion kinase (pp125 FAK ) which plays a central role in integrin-mediated signal transduction. Apart from activation by integrins, specific tyrosine phosphorylation of pp125 FAK is also induced by several growth factors (9), lysophosphatidic acid (10) and neuropeptides (11,12), suggesting that several diverse signaling pathways may converge at this point. The N-terminal sequence of pp125 FAK can bind synthetic peptides derived from several ␤-integrin cytoplasmic domains, although an in vivo association has not yet been demonstrated (2); indeed, an indirect linkage via talin (13) has recently been proposed. Within the C terminus of pp125 FAK , a focal adhesion targeting sequence is both necessary and sufficient to localize pp125 FAK to focal adhesions (14). The cytoskeletal protein, paxillin, also associates with pp125 FAK through a C-terminal sequence of pp125 FAK distinct from the focal adhesion targeting domain (15,16). Upon integrin clustering, pp125 FAK is autophosphorylated on tyrosine 397 (17), and its tyrosine kinase activity is enhanced, which may then lead to association of pp125 FAK with SH2 domain-containing signaling proteins such as Src, GRB2, phosphatidylinositol 3-kinase (PI 3-kinase), Cterminal Src kinase, and phospholipase C␥ to potentially form large signaling complexes which can, for instance, activate mitogen-activated protein kinase. Growth factors such as PDGF stimulate tyrosine phosphorylation of paxillin and pp125 FAK (9) and induce the association of PI 3-kinase with pp125 FAK in adherent Swiss 3T3 cells (18). Lysophosphatidic acid and neuropeptides such as bombesin also stimulate tyrosine phosphorylation of pp125 FAK and induce formation of focal adhesions and actin stress fibers in Swiss 3T3 cells (6,10,12). However, while associated with cytoskeletal rearrangements, a definitive role for pp125 FAK in signal transduction mechanisms that alter cytoskeletal properties remains to be determined. As discussed above, most stimuli increase pp125 FAK tyrosine phosphorylation. Surprisingly (and perhaps in contrast to its "PDGF-like" effect on cytoskeletal changes in some cells) (5), insulin stimulates the dephosphorylation of pp125 FAK in rat 1 fibroblasts (19) and in CHO cells (20). Unlike the bell-shaped curve of pp125 FAK phosphorylation in response to PDGF, pp125 FAK dephosphorylation occurs at all insulin concentrations (21). Insulin stimulation was also reported to correlate with a marked decrease in the length and number of actin stress fibers in CHO cells overexpressing insulin receptors (22). We now report that expression, in CHO cells, of an insulin receptor mutated to change lysines 164 and 582 to asparagine leads, in the absence of insulin, to decreased tyrosine phosphorylation of pp125 FAK , altered cell shape, and abrogated cell adhesion. These sites, which are putative tryptic cleavage sites (23,24), were mutated as part of an ongoing study to examine the mechanism of tryptic activation of the insulin receptor (24). The unusual phenotype of the cell line expressing the mutant insulin receptor provides further evidence for the insulin receptor as a potential regulator of cell architecture and adhesion. EXPERIMENTAL PROCEDURES Materials-Human insulin receptor containing Lys 164 3 Asn, Lys 582 3 Asn, or K164N/K582N (2N) substitutions were generated using polymerase chain reaction-based mutagenesis by Dr. L. Macaulay (CSIRO, Australia). Rat tail collagen (type I) was kindly provided by Dr. Bob Whitehead (Ludwig Institute, Melbourne, Australia). The mouse myeloid cell line (FDC-P1) was generously provided by Dr. A. Harris (Walter and Eliza Hall Institute, Melbourne, Australia). Monoclonal antibodies against the human insulin receptor, CT-1, and 18-44 were kindly provided by Prof. K. Siddle, University of Cambridge, UK. Antibodies 2A7 (monoclonal) and BC3 (polyclonal) against pp125 FAK were generously provided by Dr. J. T. Parsons, University of Virginia, Charlottesville. The mouse anti-hamster integrin ␣ 5 (clone PB1) and antihamster integrin ␤ 1 (clone 7E2) monoclonal antibodies were generous gifts from Dr. R. Juliano, University of North Carolina, Chapel Hill. Lipofectin® and G418 (Geneticin) were purchased from Life Technologies, Inc. Plasma fibronectin was purchased from Promega, Madison, WI. Insulin kinase substrate "FYF peptide" (RRDIFETDYFRK) was purchased from Auspep, Australia. Radioactive [␥-32 P]ATP (3000 Ci/ mmol) was purchased from Bresatec, Australia. Rhodamine (TRITC)labeled phalloidin was purchased from Sigma. FITC-labeled sheep antimouse IgG used for FACS sorting and sheep anti-rabbit and anti-mouse IgG F(ab) 2 fraction affinity isolated FITC-conjugated antibodies for FACS analysis were purchased from Silenius, Australia. The polyclonal antibody against a peptide corresponding to the amino acids 1223-1242 mapping at the C terminus of IRS-1 (clone C-20) was purchased from Santa Cruz Biotechnology, Santa Cruz, CA. The monoclonal anti-human vinculin (clone h-VIN1) was purchased from Sigma, and monoclonal antibody against paxillin was purchased from Transduction Laboratories, Lexington, KY. The following anti-integrin antibodies were purchased from Chemicon International Inc., Temecula, CA: mouse anti-human ␣ v monoclonal, rabbit anti-␣ 3 affinity purified, and rabbit anti-␤ 5 polyclonals. The monoclonal anti-rat LFA-1 ␤ 2 chain was purchased from Seikagaku Corporation, Toyko, Japan. Anti-phosphotyrosine antibodies conjugated to horse radish peroxidase (HRP) were purchased either from ICN Biomedicals, Cleveland, OH (clone PY20) or Upstate Biotechnology Inc., Lake Placid, NY (clone 4G10). Secondary antibodies used for immunoblotting and conjugated to HRP were purchased from Dako Corp., Carpinteria, CA. Enhanced chemiluminescence reagents were from Dupont NEN. All other reagents were of the highest grade available. Site-directed Mutagenesis-A Lys 164 3 Asn mutant receptor cDNA was constructed by inserting a mutagenic oligonucleotide (generated by a polymerase chain reaction, encoding the Lys 164 3 Asn mutation between the KpnI and EcoRI restriction sites of the human insulin receptor cDNA) into a KpnI-EcoRI digest of the insulin receptor expression vector pET (25). For mutagenesis of Lys 582 3 Asn, a mutant oligonucleotide was generated by a polymerase chain reaction between the AccI and BamHI sites of the insulin receptor cDNA, and then inserted into the AccI-BamHI digest of a modified pET insulin receptor expression vector. The single mutations were combined to produce the 2N mutant by inserting the EcoRI-XbaI fragment from Asn 582 into the Asn 164 construct. All constructs were sequenced around the insertion, subjected to extensive restriction analysis, and shown to be correct. Cell Culture-CHO cell lines were maintained in ␣-modified Eagle's medium (␣-MEM) containing 10% v/v fetal calf serum (FCS). For immunofluorescence, cells were cultured on glass coverslips to 80% confluency. When required cells were serum-starved for 18 h with ␣-MEM containing either 0.5% v/v FCS or 0.1% w/v bovine serum albumin (BSA). Stable Expression of cDNA in CHO Cells-CHO cell lines expressing the 2N (K164N/K582N) insulin receptor (and single point mutants) were obtained by co-transfection of 10 g of plasmid DNA and 2 g of pSVNeo DNA using Lipofectin®. Transfected CHO cell lines were maintained in medium containing 800 g/ml G418 sulfate to select for stable colonies. Positive insulin receptor expressing clones were identified by enzyme-linked immunosorbent assay as described by Clark et al. (24). The only variation to the described method (24) was the addition of 10 mg/ml BSA after coating the microtiter plates with CT-1, to block nonspecific binding. Cell lines expressing high levels of mutant receptors were further selected by fluorescence-activated cell sorting (FACS) as described previously (26), using nonbiotinylated antibody 18 -44 followed by FITC-labeled sheep anti-mouse IgG. In some experiments, cell sorting with simultaneous single cell cloning was performed. Several independently derived clones were analyzed. 125 I-Insulin-binding Assay-CHO cell lines were plated in 24-well plates and incubated in Earle's balanced salt solution, 25 mM Hepes, pH 7.4, and 0.1% w/v BSA containing approximately 20,000 cpm 125 Iinsulin in the absence or presence of increasing concentrations of unlabeled insulin (1-1000 ng/ml) for 4 h at 8°C. Cells were washed and solubilized, and bound radioactivity was measured in an autogamma spectrometer as described previously (27). Nonspecific binding was measured in the presence of 10 g/ml unlabeled insulin and was always less than 2% of total binding. Data were analyzed by the method of Scatchard (28). Receptor Autophosphorylation and Kinase Activity-CHO cells were treated with or without insulin (172 nM) and lysed, and insulin receptors were immunoprecipitated with monoclonal antibody CT-1 as described previously (27). In vitro autophosphorylation assays were carried out after addition of [␥-32 P]ATP as described previously (27). To measure peptide phosphorylation, receptor immunoprecipates were incubated in 40 l of phosphorylation buffer (50 mM Hepes, pH 7.4, 12 mM MgCl 2 , 100 M Na 3 VO 4 , 2 mM MnCl 2 , 150 mM NaCl, 0.2% v/v Triton X-100, 50 M ATP, and 1 mM EGTA), and the reaction was started by the addition of 200 M FYF peptide (RRDIFETDYFRK) and 0.4 Ci of [␥-32 P]ATP per sample for 5 min at 30°C. At this time, duplicate 10-l aliquots were spotted onto P81 Whatman filters, washed three times in 30% w/v glacial acetic acid, 0.05% w/v phosphoric acid, rinsed in 70% v/v ethanol, and dried, and radioactivity was measured in a liquid scintillation spectrometer. Blank values, in the absence of FYF peptide, were subtracted from each result. For IRS-1 immunoprecipitation, cells were treated as described by Konstantopoulos and Clark (21). For insulin receptor ␤-subunit immunoprecipitation, cells were treated as described above and immunoprecipitated with monoclonal antibody 18-44 overnight at 4°C. To detect tyrosine phosphorylation of either IRS-1 or insulin receptor ␤-subunit, membranes were probed with HRP-conjugated 4G10 (1:188). Measurements of Cell Growth Rate and Cell Attachment-CHO cell lines were seeded at 10 ϫ 10 3 cells/ml in 24-well plates and maintained in ␣-MEM containing 10% v/v FCS. Cells from individual wells were washed twice with Versene (phosphate-buffered saline containing 0.2% v/v EDTA), and attached cells were removed by incubation with 0.25% v/v trypsin/Versene for 3 min at 37°C. Cells were counted (in triplicate) in a Coulter counter every 24 h over 5 days, and simultaneously, the Coulter counter measured the mean cell volume. For attachment assays plates were left uncoated or coated with 10 g/ml fibronectin or 15 g/ml rat tail collagen. Treated or untreated plates were blocked with 2 mg/ml BSA for 1 h at 37°C and washed with phosphate-buffered saline prior to the addition of cells. Cells (10 -30 ϫ 10 3 /ml) were allowed to attach for either 2 or 24 h after plating in the presence of ␣-MEM containing 0.1% BSA. Actin Staining of CHO Cell Lines-For filamentous actin localization, cells grown on coverslips were fixed in 4% formaldehyde in Trisbuffered saline for 8 min at 22°C, rinsed in Tris-buffered saline, and permeabilized for 5 min in Tris-buffered saline containing 0.5% (v/v) Triton X-100. After washing in Tris-buffered saline, cells were incubated with rhodamine-phalloidin (0.1 g/ml) for 20 min at 22°C in a dark humidity chamber. After the final washes, the coverslips were mounted with a drop of Gelvatol. Cells were viewed on a LaborLux D fluorescence microscope and photographed with Kodak TMAX 400ASA film. Protein Measurements-Cellular protein concentrations were deter-mined using Coomassie Brilliant Blue dye with bovine serum albumin standards (31). Statistical Analyses-Numerical results are expressed as the mean Ϯ S.E. and statistical comparisons were performed by Student's t test for unpaired samples; p Ͻ 0.05 was taken as significant. (2N ϭ K164N/K582N) was initially difficult to obtain, despite production of stable cell lines expressing either of the individual point mutations alone (Lys 164 3 Asn or Lys 582 3 Asn). To avoid overgrowth of 2Nexpressing cells by non-2N-expressing cells, CHO.2N cells were simultaneously FACS sorted and single-cell cloned 10 days after transfection. This method provided eight independently derived clones expressing the 2N insulin receptor (up to 15 passages) which have been studied to varying degrees. One of the highest receptor-expressing clones (2N-10) was used in most of the studies, as its binding and phosphorylation characteristics (see below) were closest to those of CHO.T cells (overexpress wild-type human insulin receptors). The mutations in the external domain of the insulin receptor were of potential trypsin cleavage sites (23,24) and mutated as part of another study (24). However, the unusual phenotype of the cells expressing the double point mutation (see below) led to the analysis presented here. In preliminary experiments, treatment of the cell lines (expressing mutant insulin receptors) with trypsin did not result in significant differences from CHO.T cells. 4 Insulin-binding Characteristics-The ability of two CHO.2N cell lines (clones 10 and 11) to bind insulin is shown in Fig. 1. The insulin binding of CHO.2N-10 and CHO.2N-11 cells was lower than that of CHO.T cells (clone 10 at passage 6, 9.8%, and clone 11 at passage 6, 13.2%, compared to wild-type, 42.0%, of total 125 I-insulin added). In addition, 125 I-insulin binding to CHO.2N-10 and 2N-11 cells decreased with prolonged passage, whereas the ability of CHO.T cells to bind 125 I-insulin did not alter with passage. 125 I-insulin binding always correlated with the level of insulin receptor expression measured by enzymelinked immunosorbent assay (data not shown). CHO.2N-10 cells express fewer insulin receptors per cell compared to CHO.T cells as determined by Scatchard analysis (Table I) of insulin (0.47% of total 125 I-insulin added, see Fig. 1). The dissociation constants for insulin binding were comparable between the cell lines (Table I). Expression of 2N cDNA in CHO Cells-Expression of a double point mutant of the insulin receptor Kinase Activity of CHO.2N Insulin Receptors-The autophosphorylation of insulin-stimulated CHO.2N-10 insulin receptors (passages 6 -10) was similar to wild-type insulin receptors (Fig. 2, lower panel) as measured by immunoblotting with anti-phosphotyrosine antibodies. Autophosphorylation of CHO.2N-10 receptors was stimulated 4.4-fold by insulin compared to 5.6-fold for wild-type receptors (Table I) in an in vitro phosphorylation assay. When the insulin-stimulated kinase activity of CHO.2N-10 and wild-type receptors toward a synthetic peptide (FYF, derived from the major autophosphorylation domain of the insulin receptor tyrosine kinase) was measured, phosphorylation was stimu-lated 20.1-and 19.9-fold, respectively (Table I). Insulin-stimulated phosphorylation of pp185/IRS-1 was also comparable between CHO.T and CHO.2N-10 (passage 6 -8) cells (Fig. 2, upper panel). Cellular Morphology and Growth Characteristics-CHO.2N-10 cells (as well as four other 2N clones) showed morphological differences which disappeared upon prolonged passage. These morphological changes were not dependent on cell density. Fig. 3A is a photomicrograph of CHO.2N-10 cells at passage 7; these cells formed a heterogenous monolayer with many large, flat cells containing large, granular nuclei. In contrast, the nontransfected, parental CHO cells (Fig. 3B) or CHO.T cells (not shown) grew as a homogenous monolayer of smaller cells. At late passages (Ͼ15 passages), CHO.2N-10 cells were indistinguishable from CHO cells (not shown). This reversion in morphology strongly correlated with the loss of insulin binding (Fig. 1) and insulin receptor expression of CHO.2N-10 cells at these later passages. Rhodamine-phalloidin staining of CHO.2N-10 cells at passage 9 shows the dense formation of actin stress fibers running the length of the mutant cells (Fig. 3C). In contrast, the nontransfected CHO cells (Fig. 3D) or CHO.T cells (not shown) stain intensely close to the plasma membrane region of the cells, which appears to be absent in the CHO.2N-10 cells (Fig. 3C). The growth of CHO.2N cell lines, especially clone 10, appeared extremely slow compared to nontransfected CHO and wild-type CHO.T cells. However, the doubling time of CHO.2N-10 cells at all passages was similar to nontransfected CHO cells when corrected for the number of cells attached to dishes after 24 h (Fig. 4). In addition, there was a significant difference in the mean cell volume, with early passage CHO.2N-10 cells having a larger volume compared to CHO cells (1973 Ϯ 42 fl compared to 1231 Ϯ 42 fl, p Ͻ 0.01, n ϭ 5) correlating with the observed differences in cell size (Fig. 3, A compared to B). apparent slower cell growth of CHO.2N-10 cells was explained by their inability to attach efficiently to tissue culture dishes at early passage. After 24 h (reflects both cell attachment and cell growth) CHO.2N-10 (passage 6) cell number was 31% of added cells compared to CHO.2N-10 cells at late passage (150% of added cells) or controls (140% of added cells) (Table II). This reduced ability to adhere to the tissue culture dishes was observed with six out of the eight independently derived clones (data not shown). This was not due to differences in extracellular matrix production by early passage CHO.2N-10 cells as these cells still showed reduced attachment to matrix elaborated by CHO.T cells, and conversely CHO.T cells attached equally well to their own or a CHO.2N-10-elaborated matrix after 2 h (data not shown). In addition, early passage CHO.2N-10 cells attached less well to both fibronectin and type I collagen-coated plates after 2 h when compared to controls (Table II). Viability studies showed that the cells which did not attach to tissue culture plates were still viable (83 Ϯ 4%), as determined by trypan blue exclusion. FIG. 2. Insulin stimulates IRS-1 and insulin receptor ␤-subunit phosphorylation in CHO (a and b), CHO.T (c and d), and CHO.2N-10 (e and f) cells. Cells were incubated either in the absence Focal Adhesion Kinase (pp125 FAK ) and Other Focal Adhesion Proteins-Focal adhesions are cellular regions that, in vitro, provide contact points for cells with tissue culture dishes. They contain, among other proteins, a cytoplasmic tyrosine kinase, pp125 FAK , which is phosphorylated upon cell attachment (32). CHO.2N-10 cells at early passage have a significant reduction in tyrosine phosphorylation of pp125 FAK in the absence of insulin (42.5 Ϯ 9.7% of CHO cells, Figs. 5, panel A, and 6, upper panel). This difference held whether sample loading was corrected for either cell number (Fig. 5, panel A) or cell protein content (data not shown). It is interesting to note that expression of the wild-type insulin receptor in CHO cells (CHO.T) resulted in a consistent but not significant increase of pp125 FAK tyrosine phosphorylation (168.9 Ϯ 30.0%) compared to nontransfected CHO cells (Fig. 5, panel A). Focal adhesion kinase protein levels were comparable between CHO, CHO.T, and CHO.2N-10 cells (passages 5-8) (100 Ϯ 12%, 77.7 Ϯ 3.4%, and 77.1 Ϯ 4.9%). We also measured pp125 FAK tyrosine phosphorylation as a function of adhesion to fibronectin in all cell lines as cell attachment to fibronectin often leads to an increase in pp125 FAK phosphorylation (32). However, none of the cell lines showed a further increase in pp125 FAK phosphorylation following 2 h of plating onto fibronectin when compared to plating on noncoated dishes (data not shown). Alteration to the basal level of pp125 FAK phosphorylation could result from the action of a phosphatase, we therefore preincubated CHO.2N cells with the tyrosine phosphatase inhibitor pervanadate (1 or 10 M) for 1 h at 37°C. The basal level of pp125 FAK phosphorylation in CHO.2N cells increased significantly toward that of CHO.T cells (Fig. 5, panel B). Recently, two laboratories have shown that insulin, unlike other growth factors, stimulates dephosphorylation of pp125 FAK in cells overexpressing wild-type insulin receptors (19,20). We have also examined the effect of insulin on pp125 FAK in CHO, CHO.T, and early passage CHO.2N-10 cells. Insulin decreased pp125 FAK tyrosine phosphorylation in both CHO.T and CHO.2N-10 (passages 7 and 8) cells (Fig. 5, panels C and D) when compared to basal pp125 FAK tyrosine phosphorylation (27.9 Ϯ 7.9% and 52.2 Ϯ 10.0% of nonstimulated cells, respectively). CHO cells which express only endogenous hamster insulin receptors showed negligible dephosphorylation of pp125 FAK in response to insulin (91.0 Ϯ 16.5%, Fig. 5, panels C and D). The protein levels of pp125 FAK were not altered in response to insulin stimulation (data not shown). Results were similar whether cells were tested in complete medium or after 18 h of serum starvation. Paxillin, a structural protein, is also phosphorylated on tyrosine residues, usually (15) but not always (33,34) in parallel to pp125 FAK . In our studies, in contrast to the decrease in pp125 FAK tyrosine phosphorylation, there was no consistent change in paxillin phosphorylation (Fig. 6, middle panel) in CHO.2N-10 cells (passages 7-10). Over a number of experiments, less paxillin protein was immunoprecipitated from CHO.2N-10 cells (61.0 Ϯ 10.4%, p Ͻ 0.01, n ϭ 7) when compared to CHO and CHO.T cells. However, the ratio of paxillin phosphorylation to paxillin protein levels was similar for all a Cells were seeded at 10 ϫ 10 3 /ml in 24-well plates, and after 24 h cells were counted as described. b Cells were seeded at 30 ϫ 10 3 /ml in 24-well plates coated as described under "Experimental Procedures," and after 2 h attached cells were counted. c All results are expressed as the mean of triplicate determinations Ϯ standard error of the mean, and are representative of five separate experiments. *p Ͻ 0.05 when compared with CHO, CHO.T, and late passage CHO.2N-10 cells. cell lines. Paxillin binds to the rod domain of vinculin, another structural focal adhesion protein concentrated at the sites of cell-extracellular matrix adhesions involving integrins (35). Vinculin levels were also comparable between CHO, CHO.T, and CHO.2N-10 (passages 7 and 8) cells (Fig. 6, lower panel). In order to estimate the relative number of focal contacts for each cell line we examined the cellular distribution of vinculin by indirect immunofluorescence. Quantitation was not possible due to the gross differences in cell size between early passage CHO.2N cells and other CHO cell lines (Fig. 3, A compared to B); however, no overall difference in cellular localization of vinculin was observed (data not shown). We also attempted to estimate the number of focal adhesions present in the adherent cells by immunolocalizing the focal adhesion and membrane ruffle marker, talin. Talin has not only been shown to be indirectly linked with pp125 FAK (13) but also linked directly to the ␤ 1 cytoplasmic portion of integrin (36,37). CHO cell lines did not exhibit well developed focal contacts (also reported by Bauer et al. (38)) but rather showed an intense staining for talin around the periphery of the cell. This pattern of staining was not significantly different between CHO.2N, CHO.T, and CHO cells (data not shown). Alterations to the expression of integrins (receptors for extracellular matrix proteins) could provide an explanation for the morphological changes in CHO.2N cells. We investigated any gross changes to the level or type of integrin expression in CHO.2N cells by FACS analysis. CHO cells endogenously express significant levels of ␣ 5 and ␤ 1 integrins, and lower levels of ␣ 3 , ␣ v , and ␤ 5 integrins. 3 In general, no change to this pattern of expression was observed in CHO.2N cells compared with the other cell lines; however, there was a trend for an increase in the mean fluorescence of ␣ 3 integrins (319.8 Ϯ 91.5%, n ϭ 5 where FITC alone is corrected to 100%) compared to either CHO.T or CHO cells (181.3 Ϯ 4.5% and 175.4 Ϯ 16.2%, respectively, n ϭ 5) (Fig. 7), but this did not reach statistical significance due to the larger standard error of the CHO.2N cells as a result of their heterogeneity. It was interesting to note the 2-3-fold increase (p Ͻ 0.05, n ϭ 3) in expression of ␤ 5 integrins in cell lines which overexpress insulin receptor constructs (CHO.T and CHO.2N) compared to CHO cells (Fig. 7). DISCUSSION We have shown that mutation of two potential tryptic cleavage sites (23) the insulin receptor ␣-subunit results in a cell line with altered morphology associated with an increase in cell size, a decrease in cell adhesiveness, a decrease in protein levels of paxillin, and a decrease in tyrosine phosphorylation of focal adhesion kinase. These features correlate with expression of the mutant 2N insulin receptor. The cells revert to normal CHO phenotype at later passages, at which time insulin binding and receptor expression returns to that of nontransfected CHO cells, presumably because the cells lose the capacity to express the mutant insulin receptor. However, decreased CHO.2N cell adhesion was not associated with altered insulin receptor functions, including insulin binding, receptor kinase activity, or IRS-1 phosphorylation, nor was it associated with changes to the structural focal adhesion protein, vinculin, or gross alterations in integrin expression. While a complete survey of focal adhesion and cytoskeletal proteins has not been conducted, the reduction of pp125 FAK phosphorylation and of paxillin protein in CHO.2N cells and the known correlation of these proteins with focal contact formation (14,39,40) suggest changes to focal contacts formed by CHO.2N cells may impair their ability to interact initially with a substratum. Overall, these results support a role for the insulin receptor in modulating cytoskeletal interactions and cell adhesion. The signals generated as a result of pp125 FAK activation are currently unknown but it is apparent that autophosphorylation of pp125 FAK results in co-association of signaling molecules such as Src, PI 3-kinase, and GRB2, in addition to proteins such as paxillin which, in turn, binds C-terminal Src kinase and vinculin (2,8). Presumably generation of signals via this "facilitator complex" could maintain or change cell architecture via cytoskeletal rearrangements; a defect in this signaling system may result in the altered cell phenotype observed in CHO.2N cells. Cytoskeletal changes have already been reported following overexpression of some signaling proteins, including mitogen-activated protein kinase (41) and C-terminal Src kinase (42). Thus, the underphosphorylation of pp125 FAK in CHO.2N-10 cells may contribute to their altered morphology and adhesion. PI 3-kinase is a major signaling molecule activated following growth factor stimulation and has been implicated in pathways leading to cytoskeletal rearrangements, possibly upstream of Rac (5,43). The interaction of PI 3-kinase with pp125 FAK can be stimulated in vivo by either PDGF (18) or cell adhesion (44) concomitant with pp125 FAK activation; however, no increase in pp125 FAK -associated PI 3-kinase was seen in insulin-stimulated cells (18). Recently, it has been shown that blocking PI 3-kinase activity with wortmannin inhibits PDGF-stimulated pp125 FAK phosphorylation (45). While wortmannin and mutated p85 subunits of PI 3-kinase block insulin-stimulated membrane ruffling (5), our preliminary data show that wortmannin (100 nM) had no effect on the ability of insulin to stimulate dephosphorylation of pp125 FAK . Surprisingly in CHO.2N-10 cells, in the absence of insulin, a consistent increase in pp125 FAK phosphorylation was observed after 10 min of incubation with wortmannin, implying some interaction between PI 3-kinase and pp125 FAK in these cells. 4 Further investigations are under way to examine this linkage. A decrease in phosphorylation of pp125 FAK could result from activation of a tyrosine phosphatase. A number of tyrosine phosphatases have been proposed to play a role, both positive and negative, in insulin action. Of these, LAR (transmembrane) PTP1B and Syp (cytosolic) appear in the literature most commonly linked with insulin signaling and modulation of their activity, by various means, may either enhance (LAR (46) and PTP1B (47)) or block (Syp (20)) insulin action. Phosphatases have also been linked with cytoskeletal activities and, for instance, changes in cell density (which affect the cytoskeleton) also lead to changes in the level of LAR, PTP1B, and Syp (48). Syp, with two phosphotyrosine SH2 binding domains, has been shown to bind directly to both phosphorylated insulin receptor and IRS-1 (49) and, when overexpressed in CHO cells, a dominant negative Syp mutant induced a significant increase in basal pp125 FAK phosphorylation (20). Furthermore, LAR has been localized to focal adhesions with a possible involvement in focal adhesion disassembly (50). Our initial data which shows that the tyrosine phosphatase inhibitor pervanadate can increase the basal phosphorylation of pp125 FAK in CHO.2N cells is suggestive of phosphatase involvement. A study to isolate putative "pp125 FAK " phosphatases is in progress. The structural focal adhesion protein, paxillin, is closely linked with pp125 FAK via an association site in the C terminus of pp125 FAK . It is thought that paxillin is an in vivo substrate of pp125 FAK (2), and the phosphorylation of these two proteins usually increases and decreases in concert. However, in CHO.2N-10 cells, despite decreased basal phosphorylation of pp125 FAK , there was no parallel decrease in paxillin phosphorylation demonstrating that the two events are not always linked. A similar result was observed in phorbol 12-myristate 13-acetate down-regulated cells (33) and pp125 FAK -deficient mice (34). Little is known about the overall mechanism for focal contact formation and the contribution of various structural proteins such as paxillin. In this context, our observation that the level of paxillin protein is decreased by 40% in CHO.2N-10 cells could explain (at least in part) the decreased ability of these cells to adhere, despite no alteration to the level of vinculin. The mechanism whereby a double point mutation (K164N/ K582N) in the external domain of the insulin receptor leads to the observed cell phenotype is unknown. One possibility is that the mutation results in a conformational change in the receptor's cytoplasmic domain such that a signal transduction pathway involved in cell architecture is modulated in the absence of insulin. Truncation of the insulin receptor C terminus and more specifically, mutation of tyrosines 1328 and 1334 in this region abolished the insulin responsive dephosphorylation of pp125 FAK , suggesting that the C terminus and in particular phosphorylation of these two tyrosine residues is critical for insulin effects on pp125 FAK phosphorylation (19). In addition, this region of the receptor and in particular Tyr 1328 and Tyr 1334 has been linked to Syp activation by insulin (51), and also to Shc (52), p85 subunit of PI 3-kinase (53), and Grb10 (54) association. Previous studies have demonstrated that insulin binding to its receptor induces a conformational change in the C terminus of the ␤-subunit prior to and independent of autophosphorylation (55). Again mutation of residues Tyr 1328 and Tyr 1334 of the insulin receptor alters this insulin-induced conformational change (55). Taken together these data suggest that modulation of the insulin receptor ␣-subunit (by binding insulin) can influence the conformational state of the C terminus and that these C-terminal sequences in turn may play a role in signal transmission. Perhaps the 2N mutation has also resulted in a C-terminal conformational change altering a pathway (containing pp125 FAK ) involved with cell morphology. As antibodies are available that recognize different conformational states of the receptor C terminus these hypotheses could be tested. Finally, the recent demonstration by Richardson and Parsons (40) that overexpression of the pp125 FAK -related non-kinase (pp41/43 FRNK ) blocks pp125 FAK tyrosine phosphorylation and inhibits cell spreading supports the notion that alterations to pp125 FAK can be linked with changes to cytoskeletal interactions. Further studies will be required to elucidate the mechanisms involved in these changes in CHO.2N cells; however, our results linking expression of a mutant insulin receptor with altered cell phenotype add support to a role for the insulin receptor in organization of the cell cytoskeleton.	2018-04-03T05:42:00.284Z	1996-11-15T00:00:00.000	{ "year": 1996, "sha1": "1631e6d3a011d4bb66e0280651587a8ee01d81c1", "oa_license": "CCBY", "oa_url": "https://doi.org/10.1074/jbc.271.46.28960", "oa_status": "HYBRID", "pdf_src": "Highwire", "pdf_hash": "fabe8cd7311296a4dc6d8d900bf4f36decc86e41", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
247208342	pes2o/s2orc	v3-fos-license	A Rare Case of Aseptic Herpes Simplex Virus-1 Meningitis in a Patient with Recent COVID-19 The aim of presenting this case was to show the difficulties in making the differential diagnosis of viral meningitis during the COVID pandemic situation. We report a case of a young man with clinical features of viral meningitis and with epidemiological history of COVID-19 in his family. The patient complained of fever, headache, photophobia, nausea, myalgia, and fatigue. He gave a history of diarrhea and vomiting two weeks before admission and close relatives with COVID-19. The neurological examination revealed a meningeal irritation syndrome. The diagnostic tests we performed were as follows: nasopharyngeal swab and PCR of cerebrospinal fluid (CSF) for SARS CoV-2, computed tomography scan of the head, general CSF examination, viral tests, and microbiology of CSF, enzyme-linked immunosorbent assay (ELISA) IgM and IgG. The results were consistent with viral meningitis due to HSV-1 in simultaneously found high titres of plasma SARS CoV-2 specific IgA and SARS CoV-2 specific IgG and active viral serum infection for CMV and EBV. INTRODUCTION Aseptic meningitis is an illness characterized by serous inflammation of the linings of the brain (i.e., meninges), usually with an accompanying mononuclear pleocytosis. Clinical manifestations vary, with headache and fever predominating, stiff neck, photophobia, drowsiness, myalgias, chills, sore throat, abdominal pain, nausea, and vomiting. The illness is usually mild and runs its course without treatment; however, some cases can be severe and life-threatening. Immunosuppressed patients are largely at a high risk for developing the condition. Aseptic meningitis syndrome is not caused by pyogenic bacteria. Although it is usually caused by certain viruses, it has a number of other etiologies as well, both infectious and non-infectious. Hence, the term aseptic meningitis is no longer synonymous with viral meningitis, although the two are still often used nterchangeably. Viral causes include the following: enteroviruses (coxsackievirus, echovirus, poliovirus), herpes simplex virus types 1 and 2 (HSV-1, HSV-2), varicella-zoster virus, arboviruses, the Epstein-Barr virus, HIV, influenza virus types A and B, the mumps virus, lymphocytic choriomeningitis virus (LCMV), and rabies. Mollaret meningitis is a recurrent disorder whose causative agent remains unknown. However, recent data suggest that herpes simplex virus (HSV-2 and, less frequently, HSV-1) may cause some if not most cases. 1 It is characterized by repeated episodes of meningitis, typically lasting two to five days, occurring weeks to years apart. 2 Viral meningitis is a relatively common disorder. The incidence of aseptic meningitis has been reported as 11 per 100,000 person-years, compared with a rate of 8.6 per 100,000 for bacterial meningitis. 1 No specific treatment exists for most of the viruses that cause meningitis; therefore, management, for the most part, is supportive and includes analgesics, antinauseant medications, intravenous fluids, and prevention and treatment of complications. Given the potential for serious neurological morbidity and the persistently high mortality rates of bacterial meningitis, a rapid institution of antibiotic coverage is essential when the diagnosis of bacterial meningitis is suspected. A third-generation intravenous cephalosporin is a customary choice. Effective antiviral therapy is available against HSV-1, varicella, and cytomegalovirus. In immunosuppressed patients, long-term therapy may be necessary. Acyclovir is recommended for immunocompetent hosts with HSV-2 meningitis and a primary genital herpes infection. In patients with Mollaret meningitis, acyclovir (intravenous or oral) or valacyclovir (oral only) are worthy of consideration for both therapy and prophylaxis. In general, corticosteroids are avoided in aseptic meningitis because of their inhibitory effects on host immune responses. Occasionally, glucocorticoids, such as dexamethasone, are useful when meningitis is associated with signs of increased intracranial pressure. 1 Of the three herpesviruses that can remain latent in the host for a lifetime, herpes simplex virus 1 (HSV-1) is more associated with encephalitis, whereas HSV-2 and varicella-zoster virus (VZV) cause meningitis. These viruses cause primary infection of mucocutaneous surfaces, establish latency in the peripheral sensory ganglia through retrograde transportation, and reactivate periodically with anterograde transmission to the nerve endings and mucocutaneous surfaces. HSVs (HSV-1, HSV-2, VZV) enter the brain mainly via the peripheral and cranial nerves. Once inside the brain, the virus spreads through the subarachnoid space in cerebrospinal fluid (CSF), and its rapid replication can overcome the host defences. Inflammatory WBCs such as lymphocytes accumulate and target the virus. Consequently, inflammatory cytokines such as interleukin (IL)-1β, IL-6, and tumor necrosis factor-α are released. Inflammatory responses can increase the permeability of the blood-brain barrier, allowing the entrance of circulating immunoglobulins. 3 Novel coronavirus (SARS-Coronavirus-2; SARS-CoV-2) emerged in December 2019 in Wuhan, China, and has become a global health crisis. The SARS-CoV-2 virus affects mainly the respiratory system, although other organs are involved. 4 Symptoms are associated with lower respiratory tract infection, including fever, dry cough, and shortness of breath. Also, headache, dizziness, generalized weakness, vomiting, and diarrhea may occur. 5 The neurological complications described in SARS-CoV-2 are encephalitis and meningitis, acute necrotizing encephalopathy, acute transversal myelitis, cerebrovascular pathology, myopathy, and Guillain Barre syndrome. 6 Infection with SARS-CoV-2 is mediated by binding be-tween the receptor-binding domain of viral thorns and the cellular angiotensin-converting enzyme-2 (ACE2) receptor. ACE2 is expressed in many tissues (lungs, heart, kidneys, CNS, intestines, testicles), which suggests the emergence of multiorgan complications. 7 The T-cell response is initiated by antigen presentation by dendritic cells and macrophages. CD4+ and CD8+ T cells play a key role. CD4+ T cells activate B cells to stimulate the production of a virus-specific antibody, while CD8+ T cells can kill virus-infected cells. 8,9 CASE REPORT A 23-year-old man without any concomitant diseases presented to the emergency room with a 3-day history of headache, photophobia, nausea, myalgia, fever, and fatigue. Painkillers had no effect on the symptoms. A CT scan was performed 3 days before admission was without any pathological findings. The patient had complained of diarrhea and vomiting two weeks before admission, physical examination was performed by a general practitioner, without additional examinations. Additional epidemiological history was obtained. The patient's father died with acute respiratory distress syndrome due to COVID-19 2 months prior to the event and his grandmother was hospitalized in ICU with COVID-19 acute respiratory syndrome. The patient presented a negative result of the ambulatory nasopharyngeal swab PCR -SARS-COV-2 test. The physical examination was normal except for the high temperature of 37.7°C. The neurological examination revealed a meningeal irritation syndrome with neck stiffness, positive Brudzinski's and Kernig's signs as well as signs of altered consciousness -the patient was oriented but drowsy. Diagnostic and therapeutic management On admission, a chest X-ray was performed with a normal result. Nasopharyngeal swab for PCR-SARS-COV-2 was performed with a negative result. Blood counts also remained within the normal range, with only a slight increase in the percentage of leukocytes relative to lymphocytes, with a normal total number of white blood cells (WBC 8.3 G/l; Lym 0.5 G/l; Lym% 6.8%; GRAN 7.6 G/l; GRAN% 90%). CRP, D-dimer, and ferritin values were also within the reference ranges. Brain CT angiography and brain magnetic resonance tomography were also performed without any abnormalities. The cerebrospinal fluid leaked under normal pressure, clear, colourless. The general CSF examination was normal and the microbiological culture remained sterile. The CSF SARS-CoV-2 PCR was negative. Positive antibody titres were detected for EBV IgG (+) and IgM (+), CMV IgM (+) IgG (+), VZV IgM (+), HSV-1 IgM (+) IgG (+) in serum and CSF. Immunoglobulins are glycoproteins that act as antibodies in the body. IgM antibodies appear in the serum between 7-12 days after infection, and IgG antibodies appear in the serum after 14 days of meeting with the infectious agent. The presence of both IgM and IgG in the serum may mean that the patient is in the active phase of the infection or the recovery period. The presence of antibodies in the CSF indicates a neuroinfection. Because The patient was treated with antibiotic therapy (intravenous ceftriaxone 2×2 g), antiviral therapy (acyclovir 400 mg 5 times daily), antiedema therapy (mannitol 4×125 ml intravenously, and dexamethasone 2×8 mg intravenously), symptomatic therapy with antipyretic and analgesic medications. The patient was discharged fully recovered after 10 days of hospital treatment. Therapy with acyclovir was prescribed for 10 days. DISCUSSION The first reported case of SARS CoV-2 meningitis and encephalitis was described by Moriguchi et al. 10 in May 2020. They described a 24-year-old man with typical symptoms of meningoencephalitis with negative nasopharyngeal SARS CoV-2 PCR and positive CSF PCR, abnormalities consistent with encephalitis on brain MRI. 10 The first differential diagnosis we suspected was SARS-CoV-2 associated meningitis. Moreover, both epidemiological history and gastrointestinal complaints supported this theory. The results of nasopharyngeal and CSF SARS-CoV-2 PCR were not consistent with this suggestion. But serum ELISA SARS CoV-2 confirmed that our patient was "immune" and had undergone COVID-19 recently. As a result of severe inflammation, it is possible to reach an "exhausted immune system". T-cell exhaustion is term characterizing the progressive loss of T-cell functions and can culminate in the physical deletion of the responding cells. Exhaustion is well-defined during chronic lymphocytic choriomeningitis virus infection, severe viral infections, and commonly develops under conditions of antigen-persistence, which occur following many chronic infections that are of significant public health concern including hepatitis B virus, hepatitis C virus, and human immunodeficiency virus infections, as well as during tumor outgrowth. 11 This makes the body susceptible to many other infectious agents. We tend to assume that this is exactly what happened in the clinical case presented above. To date, numerous data have been reviewed and presented on how a decline in the immune system can lead to more serious manifestations of SARS-CoV-2 infection. But let's assume the opposite -is it possible that this viral infection contributes to the clinical manifestation of other infectious diseases? To date, no report has denied or confirmed this two-way link. The question remains open for further research. We performed a wide viral assessment in the patient's serum and CSF. Despite its high cost, polymerase chain reaction (PCR) can rapidly and accurately detect EV, HSV, VZV, EBV, and SARS-CoV-2. The results proved the presence of HSV-1 in CSF and positive serum IgM CMV and EBV. HSV-1 meningitis is extremely rare. Encephalitis is the most frequent clinical central nervous system (CNS) manifestation of HSV-1. HSV-2 typically causes meningitis. There are two main types of HSV CNS infection, primary and recurrent. In more than 70% of cases, HSV meningoencephalitis is caused by the reactivation of a latent virus in individuals who have previously been infected. Once reactivation has occurred, viral particles are transported via the anterograde axonal transport via the olfactory and trigeminal pathways to the CNS. In 30% of cases, HSV meningoencephalitis is caused by a primary infection, and the virus also reaches the CNS by the olfactory and trigeminal nerves. In immunocompetent hosts, HSV meningoencephalitis affects brain regions such as the limbic system, mesial temporal, and frontal regions (amygdala, hippocampus, parahippocampal gyrus, temporal uncus, insula, and cingulate gyrus) earlier and most severely. The putamen and basal ganglia are usually unaffected. 12 Poesy Payal et al. report a clinical case of a 47-year-old male, an alcoholic with a previous history of tuberculous lymphadenitis, who presented with headache, fever, and altered speech. CSF analysis proved positive PCR HSV-1 and Cryptococcus neoformans. 13 Somayeh Azadfar et al. evaluated HSV meningitis in 45 children in Gorgan province, Iran. Forty-five cerebrospinal fluid samples taken from children with negative bacterial culture results were tested for viral, biochemical, and cytological assays. DNA extraction and PCR were performed. HSV-1 was detected in 4 (8.8%) samples without any HSV-2 infections. Their conclusion was that the distribution of HSV types in children with meningitis in Gorgan province was predominantly type 1 compared with type 2, which was reported more in other areas. 14 HSV-1 meningitis is extremely rare. In most reported cases, meningitis is due to coinfection with HSV-1 and other agent or it is found in patients with immunodeficiency. The treatment obligatory includes antiviral medications. In our case, both key findings are the COVID-19 "terrain" and rare HSV-1 cause for meningitis.	2022-03-03T16:21:32.160Z	2022-02-28T00:00:00.000	{ "year": 2022, "sha1": "bcd07acb8b1e484f89a3997a38ee628852ba9546", "oa_license": "CCBY", "oa_url": "https://foliamedica.bg/article/59677/download/pdf/", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "d9135c99acbac7864927f29d4b7c4ef38e23aa27", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
244799302	pes2o/s2orc	v3-fos-license	Structural studies of the sol-gel glasses with copper selenide nanoparticles by SANS technique The sol-gel silica glasses doped with copper selenide nanoparticles were studied through small-angle neutron scattering (SANS) and transmission electron microscopy. SANS intensities demonstrate the complicated dependence on the scattering parameter (transferred momentum within the framework of this technique), however, with no pronounced peaks throughout the full q interval. The multi-component power-law-exponential model was applied for the approximation and evaluation of the scattering objects sizes at different levels. Also, fitting to the mono-exponential fractal model was elaborated for SANS data. From the results of both simulations we derive the sizes of Cu2Se particles about 60nm and the noticeable modification of the glass matrix at the scale of the one order more (the long-range effect of particles upon the glass). The fractal dimension of the above scattering objects is varied from 1.5 through 4.2 indicating the formation of complex structure 'particle-in-glass' which may not be reduced to a simple model of single particles in solid matrix. Introduction Copper chalcogenides, among many other perspective semiconductors, reveal novel properties combining both quantum confinement effects and the plasmon resonance phenomenon [1][2][3]. The latter appeared due to elevated free carrier concentration (electrons or holes, depending type of compounds). These compounds possess variable composition and non-stoichiometry and include a number of stable and metastable phases, e.g. in Cu 2-x Se for 1≤x≤2. The value of x strongly affects the carrier concentration and all their properties. In optics, new absorption bands and IR luminescence were discovered for the chalcogenides and associated with the plasmon resonance. Research of last years has demonstrated very interesting interplay in the nanoscale effects inherent to both metals and semiconductors. In the present short communication, we investigate the structural features of copper selenide nanoparticles embedded into amorphous (glass-like) silica matrix fabricated through the sol-gel technique [4,5] using small-angle neutron scattering (SANS). This technique allows address the structural data for a wide scale of structural features in materials of different chemical nature (1-1000 nm) with no material destruction [6,7]. Experimental The fabrication procedure is based on the sol-gel technique with hydrolysis of tetraethoxysilane resulting in the glassy silica with doping of copper salt precursor with subsequent reduction of copper followed by its selenization. It has been described earlier [4,5]. An important point of this technique is the final step of copper selenization within the silica matrix when a sample is located in the sealed quartz ampoule. This protocol of Cu x Se-doped silica fabrication allow to control both concentration of copper selenide inside and Cu/Se ratio. The latter determines the phase composition of copper selenide nanoparticles and optical features of the semiconductor nanoparticles. The size of particles appears to be in the range of tens of nanometers. They are basically isolated and can form some aggregates at the high doping concentration [6]. The transmission electron microscopy (TEM) study was carried out through the "replica with extraction" procedure. A thin carbon film (10-20 nm) was evaporated onto a freshly etched (diluted HF) surface of a sample followed by the carbon film detaching in water and transfer it to a TEM supporting grid. SANS measurements were performed on the YUMO spectrometer (IBR-2 high-pulse reactor, Dubna, Russia) [8]. The transferred momentum (q) interval was 0.006 -0.6 Е -1 allowing the considerable wide size range of scattering objects. The data were recorded at the room temperature, they were corrected taking into account the samples transmission and background scattering due to the support, a vanadium ethalone was used. The SasView software [9] was utilized for data processing. (1) , Parameters G 1 , G 2 , B 1 , B 2 and exponents P 1 и P 2 are ajustable in this model, R g1 and R g2 are giration radia for two structural levels. q* 1 and q* 2 are expressed as: Results and discussion The linear parts were fitted by the formula where A and B are fitting constants. This evaluation resulted in finding fractal dimensions for different structural levels of the system. The simulation results according to the model with the formulas (1-2) have shown in Fig. 1 by the curve (no raw measurement data are give here). The structural parameters of scattering objects deduced from this simulation, Rg 1 and Rg 2 , are 253 and 22 nm, respectively. Rg 2 value can be assigned to the nanoparticles while Rg 1 , which is approximately larger on the order of magnitude, may be associated with the areas around them. TEM images (Fig. 2) demonstrate clearly appearance of modifications in the matrix about the particles. However, Rg 1 is considerable larger than the observed areas in Fig. 2. In the other words, according to the SANS data, the modifications in the matrix are much wider than the closest cavities in Fig. 2. There is a long-range effect of the particles upon the glassy silica. The strong effect can be supposed from thedifference of thermal behavior of SiO 2 and Cu 2 Se under cooling down from the maximum temperature of the doped glass fabrication, 1200 о С. Copper selenide at this temperature exists in the melted state (melting point is 1113 о С) and its crystallization may provide an essential change of the particle volume leading to deformation of the glass around. The diameter of copper selenide nanoparticles assuming their sperical shape and the relation with giration radius R=(5/3) 1/2 R g can be determined about 57 nm. This value is smaller than the particles on TEM images, but the general consistence may be accepted as sufficient since the SANS model incorporate the wider range sizes for of scattering objects while TEM contrast resolves the maximum contrast ones. A further development of SANS treatment incorporating more structural levels (than two ones here) is expected to result in better agreement of the different measurement techniques. Fractal properties of the stuctures under study appear to be more complicated, and at least three different exponents may be deduced from the SANS data. The maximum value of α (~4.2) is for the lower q range that corresponds to the larger scattering objects and α~2.2 is obtained for the q range around 0.1 Å -1 . Likely, these are nanoparticles and their comparatively low fractal dimension can mean the rough particle-matrix interface. The lower α for the intermediate q range and its significant growth further indicate possible variation of the glass structure within the considerable wide region around the particles. Conclusion Thus, the presented SANS study of the silica sol-gel glasses doped with copper selenide nanoparticles has evidenced the formation of the complex scattering objects at the nanometer scale. The good fitting of the experimental data was shown to the model which include different structural levels with Gunier and Porod parts. Fractal properties of the material from these SANS data were also analysed. The Cu 2 Se nanoparticles of the average size about 60 nm are principal constituent of these glasses. They modify the closest environment of the silica matrix and the also the long-range effect upon its structure occurs in accordance with the model. The fractaldimension of the particle-matrix system is varied from 1.5 to 4.2 corresponding the complicated ierface and the modified silica matrix. concentration. The right sample is close to that analyzed by SANS above.	2021-12-03T02:15:44.260Z	2021-12-02T00:00:00.000	{ "year": 2021, "sha1": "75520d17f1459bfe59ffa19de6d0164fd57f1cf8", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "75520d17f1459bfe59ffa19de6d0164fd57f1cf8", "s2fieldsofstudy": [ "Materials Science" ], "extfieldsofstudy": [ "Physics" ] }
235311319	pes2o/s2orc	v3-fos-license	COVID-19 and neuropsychiatric disorders: Common links and extended networks Coronavirus disease 2019 (COVID-19), caused by the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has posed a grave threat to humanity since December, 2019. Firstly observed in Wuhan, China, COVID-19 has turned into a global threat. According to World Health Organization (WHO), “globally, as of 3:34pm CEST, 7 April 2021, there have been 132,046,206 confi rmed cases of COVID-19, including 2,867,242 deaths. The number of new deaths also increased by 11% compared to last week, with over 71 000 new deaths reported. As of 5 April 2021, a total of 604,032,357 vaccine doses has been administered” [1]. However, the vaccine against SARS-CoV-2 is in its greenhorn to serve the world and we have to wait for any other alternative capable of withstanding the progression of COVID-19 pathogenesis [2]. In addition to multiple physiological anomalies, the mental health of the COVID-19 sufferers also faces deplorable condition [3]. In other words, neuro-psychiatric abnormality coincides with the physiological abnormalities of the COVID-19 patients [4]. Though treatment strategies had been targeted towards amelioration of physiological states of the COVID-19 patients, psychological support, research and development and guidelines as well as associated regulations had received little attention. Thus, the present articles had been aimed at delving out the neuropsychiatric disturbances of the COVID-19 patients followed by their tentative remedies. Care had been taken to grasp information from the most recently available data. At the same time, constructive criticism of the already utilized practice and necessary suggestions has been put forward. Introduction Coronavirus disease 2019 (COVID-19), caused by the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has posed a grave threat to humanity since December, 2019. Firstly observed in Wuhan, China, COVID-19 has turned into a global threat. According to World Health Organization (WHO), "globally, as of 3:34pm CEST, 7 April 2021, there have been 132,046,206 confi rmed cases of COVID-19, including 2,867,242 deaths. The number of new deaths also increased by 11% compared to last week, with over 71 000 new deaths reported. As of 5 April 2021, a total of 604,032,357 vaccine doses has been administered" [1]. However, the vaccine against SARS-CoV-2 is in its greenhorn to serve the world and we have to wait for any other alternative capable of withstanding the progression of COVID-19 pathogenesis [2]. In addition to multiple physiological anomalies, the mental health of the COVID-19 sufferers also faces deplorable condition [3]. In other words, neuro-psychiatric abnormality coincides with the physiological abnormalities of the COVID-19 patients [4]. Though treatment strategies had been targeted towards amelioration of physiological states of the COVID-19 patients, psychological support, research and development and guidelines as well as associated regulations had received little attention. Thus, the present articles had been aimed at delving out the neuropsychiatric disturbances of the COVID-19 patients followed by their tentative remedies. Care had been taken to grasp information from the most recently available data. At the same time, constructive criticism of the already utilized practice and necessary suggestions has been put forward. COVID-19 and neuropsychiatric disorders: common links and extended pathophysiological networks Virus mediated respiratory tract infection leads towards multi-organ and multi-system complications including the Central Nervous System (CNS) [5]. SARS-CoV-2 utilizes the olfactory nerve and bulb to enter the CNS where they replicate in the neurons and cause infl ammation and demyelination. The virus spreads throughout the brain and Cerebrospinal Fluid (CSF) within a week [5]. Olfactory and gustatory problems (ageusia, anosmia and hyposmia) have been observed in about 50% of the COVID-19 patients [7]. Besides, the SARS-CoV-2 uses the angiotensin-converting enzyme 2 (ACE2) receptor of the oro-nasal mucosa to enter the CNS and initiate a cytokine storm that culminates into neuroinfl ammation and immune-mediated infl ammation that becomes manifested as convulsions and delirium [8]. Thus, CNS modulating effects of Citation: Rahman SARS-CoV-2 involving neuropsychiatric abnormalities remains a grave concern of COVID-19 [9,10]. Those already suffering from neurological and psychological complications become much vulnerable towards SARS-CoV-2 infection and the heightened risk warrants extensive caregiving and appropriate medico-therapy [11]. Alarmingly, the aged populace and those suffering from neurodegenerative diseases like Alzheimer's disease (AD) pose extra threat upon the already existing neuro-psychiatric disorders [11]. Neuropsychiatric symptoms associated with COVID-19 can be clustered into three groups [12] A. Olfactory symptoms: anosmia and hypogeusia. Acute neuropsychiatric problems associated with CO-VID-19 Delirium: Delirium is the most notable acute neuropsychiatric disorder among the aged COVID-19 patients especially those suffering from dementia [13]. Delirium has been observed in about 15% of the COVID-19 sufferers [6]. Behavioral alteration and poor cognitive performance associate with delirium in the COVID-19 sufferers [14]. Conclusion As COVID-19 has been paralyzing the global healthcare, psychological, economical, educational, societal and political situation harshly, immediate measures against this pandemic are inevitable. However, human endeavor since the present time, seems meager with respect to the global crisis. Thus, following the mitigations strategies and providing psychosocial support to the already affected COVID-19 patients seem apt. At the same time, the search for effective vaccine and medico-therapy against SARS-CoV-2 remains promising.	2021-06-03T07:59:37.263Z	2021-04-29T00:00:00.000	{ "year": 2021, "sha1": "3c6a56d016b1e97b2bb515764ebd71fe25b0fba8", "oa_license": "CCBY", "oa_url": "https://www.peertechzpublications.com/articles/JNNSD-7-144.pdf", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "3c6a56d016b1e97b2bb515764ebd71fe25b0fba8", "s2fieldsofstudy": [ "Psychology" ], "extfieldsofstudy": [ "Computer Science" ] }
18791539	pes2o/s2orc	v3-fos-license	Heterosubtypic anti-avian H5N1 influenza antibodies in intravenous immunoglobulins from globally separate populations protect against H5N1 infection in cell culture. With antigenically novel epidemic and pandemic influenza strains persistently on the horizon it is of fundamental importance that we understand whether heterosubtypic antibodies gained from exposures to circulating human influenzas exist and can protect against emerging novel strains. Our studies of IVIG obtained from an infection-naive population (Australian) enabled us to reveal heterosubtypic influenza antibodies that cross react with H5N1. We now expand those findings for an Australian donor population to include IVIG formulations from a variety of northern hemisphere populations. Examination of IVIGs from European and South East-Asian (Malaysian) blood donor populations further reveal heterosubtypic antibodies to H5N1 in humans from different global regions. Importantly these protect against highly pathogenic avian H5N1 infection in vitro, albeit at low titres of inhibition. Although there were qualitative and quantitative differences in binding and protection between globally different formulations, the heterosubtypic antibody activities for the respective IVIGs were in general quite similar. Of particular note because of the relative geographic proximity to the epicentre of H5N1 and the majority of human infections, was the similarity in the antibody binding responses between IVIGs from the Malayan peninsula, Europe and Australia. These findings highlight the value of employing IVIGs for the study of herd immunity, and particularly heterosubtypic antibody responses to viral antigens such as those conserved between circulating human influenzas and emerging influenza strains such as H5N1. They also open a window into a somewhat ill defined arena of antibody immunity, namely heterosubtypic immunity. INTRODUCTION Intravenous immunoglobulins (IVIGs) are widely used as highly-efficacious therapies to treat a broad range of conditions (Simon and Spath, 2003). Because IVIGs are highly-purified IgG products prepared from the plasmas of tens of thousands of blood donors they effectively reflect the herd immunity of their respective adult blood donor populations, and a sampling of the spectrum of antibodies to common infectious agents, such as influenza viruses. We earlier investigated whether Australian blood donorsuniquely geographically separated from the rest of the world on an island continent and naive to H5N1 exposure -may contain heterosubtypic antibodies, which crossreact with H5N1. Indeed, we provided evidence for heterosubtypic antibodies that bound to many of the proteins of H5N1, including both surface as well as internal proteins (Lynch et al, 2008). There are considerable uncertainties that surround the current expansion of circulating H1N1, H5N1 and H9N2 strains and their potential to evolve and re-assort to feed highly-pathogenic strains capable of human-to-human pandemic transmission. Consequently, the issue of whether there is a level of heterosubtypic antibody protection in the community at large to protect against newly emergent influenzas is of considerable importance. In this study we assessed whether intravenous immunoglobulins (IVIGs) from a range of different regions across the globe contain heterosubtypic antibodies reactive to avian H5N1 influenza. Utilizing IVIGs as a gauge of the herd IgG immunity of adult populations we determined whether there are heterosubtypic antibodies to H5N1 in IVIGs prepared from European and South East Asian (Malaysian) blood donors as we have found with Australian-derived IVIG. This study was designed to; i) assess the herd antibody immunity to a novel influenza, such as H5N1, ii) compare the herd immunities between geographically separate populations, and following positive affirmation, iii) profile the proteins of avian H5N1 influenza that IVIG crossreactive antibodies bind to, and of most importance, iv) determine whether heterosubtypic IVIG antibodies from different regions are capable of protecting against in vitro H5N1 infection. IVIGs The IVIGs studied were from a variety of different preparations, obtained from suppliers listed in Table 1. These highly-purified IgG preparations (≥ 96% pure) were isolated from plasma pools of up to 60,000 normal blood donors from the indicated geographical communities. IVIGs dilutions were prepared with TBS or PBS as indicated in the text. Table 2. The human viruses for this study were representative vaccine strains of previously circulating viruses, and were included in northern and southern hemisphere influenza vaccines between 1992 and 2000. These had been expanded in chicken eggs, and included two purified H1N1 and two H3N2 influenza isolates from humans obtained from the WHO laboratory, Parkville, Victoria, Australia. The avian influenza viruses examined included three H5N1 isolates, and for comparison H7N7 and H9N2 isolates. The avian viruses were grown in the allantoic sac of embryonated chicken eggs and purified in sucrose gradients. Briefly, allantoic fluid was collected from infected eggs, pooled and clarified by low speed centrifugation (10,000xg for 10min). Virus was pelleted in an ultracentrifuge (80,000xg for 60min), resuspended in 5 ml TNE buffer (0.01 M Tris, 0.001M EDTA, 0.1M NaCl, pH 7.2) and purified by ultracentrifugation through a 20-50% discontinuous sucrose density gradient (80,000xg for 3hrs). The visible band of virus was extracted, and virus pelleted in an ultracentrifuge (80,000xg for 60min) and re-suspended in TNE buffer to 1% of the original volume of allantoic fluid. The A/Texas/36/1991 strain was represented in northern hemiphere vaccines (NHV) from 1992 to 1995, and for southern hemisphere vaccines (SHV) for 1997: the A/ Beijing/32/1992NHV 1993to 1994and the A/Sydney/5/1997NHV 1997to 2000and SHV for 1998to 2000 The growth and expansion of the HPAI avian viruses was performed at the CSIRO Australian Animal Health Laboratories (AAHL), Geelong, Victoria. To permit removal from the AAHL secure area and to allow the virus to be handled safely for antibody binding and protein analyses the viruses were inactivated by exposure to 5 megarads of radiation. The purity of the viruses was accessed by their SDS-PAGE profiles (see below and Lynch et al, 2008). Infection studies All experiments utilizing live HPAI viruses were performed in the high-security laboratories at AAHL. Virus neutralization assays were done in MDCK (Madin-Darby Canine Kidney -ATCC CCL-34) cells or Vero (African green monkey kidney -ATCC CCL-81) using a modification of a standard protocol (Cottey et al, 2001). For this, 100 tissue culture infectious doses of each virus were added to two-fold dilutions of each serum or IVIG preparation in microtitre plate wells and incubated for 60min at 37°C. Cells were added and the plates incubated at 37°C for 3 days. Viral activation by trypsin was unnecessary as all the viruses used in this study have a multi-basic cleavage site in their hemagglutinin protein. The neutralizing titre of serum or IVIG was the highest dilution giving complete neutralization of the virus. Two-dimensional (2D) gel electrophoresis Proteins were separated using 11cm precast immobilized pH gradients (IPG) across pH ranges 4-7 (Biorad, Life Sciences, Hercules, Ca, USA) and 6-11 (GE Healthcare, Uppsala, Sweden). 300μg of protein was loaded for each sample. Isoelectric focusing of proteins was conducted according to published techniques, in the pH range 4-7 (Nouwens et al, 2000) and proteins were focused for a total of 35kVh. Proteins spanning pH 6-11 were cup loaded and focused as previously described (Nouwens et al, 2000) for a total of 38kvh. The IPG strips were reduced, alkylated, and the detergent was exchanged, prior to second dimension separation by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (Nouwens et al, 2000). SDS-PAGE was carried out in 11cm 8-16% (w/v) polyacrylamide/Tris-HCl gels (Bio-Rad) using a Criterion Dodeca Cell (Bio-Rad). Gels were fixed in 10% (v/v) methanol, 7% (v/v) acetic acid for 1hr, and then stained with Sypro Ruby (Molecular Probes, OR, USA) overnight. Gels were de-stained for 2hr in 10% (v/v) methanol and 7% (v/v) acetic acid and scanned using a ProXPRESS Proteomic Imaging System (PerkinElmer, MA, USA). One-dimensional SDS-PAGE and immunoblotting Viruses were subjected to SDS-solubilization and electrophoretic separation in criterion (Biorad, Life Sciences, Hercules, CA, USA) 4-16% (w/v) gradient gels (SDS-PAGE) under non-reducing conditions (Sloane et al, 2005). The relative molecular weights of influenza proteins were calculated from in-gel migration of prestained molecular weight standards (Sea Blue Plus2; Invitrogen Life Sciences). As the mobility of some prestained proteins can vary between batches as well as gel type, to account for this each batch used was compared inhouse against a panel of unstained proteins over the molecular weight range, and where necessary the assigned molecular weight were recalibrated for our tests. As indicated the SDS-PAGE gel separated viral proteins were visualised by Coomassie Blue R (Sigma-Aldrich) or silver staining (Biorad). For immunoblotting the SDS-PAGE separated viral proteins were electrotransfered onto 0.2µm Protran reg nitrocellulose (Schleicher and Schuell, Whatman, PE-Life & Analytical Sciences, Boston, MA), followed by blocking with 5% (w/v) skimmed milk powder in TBS. The antibody-binding step was performed using 1/200-1/500 dilutions of IVIG (50mg/ml), incubated for 1hr at room temperature (RT). After washing with TBS, blots were incubated at RT for 1hr with a 1/40,000 dilution of a horseradish peroxidase (HRP) goat anti-whole human antibody (Sigma-Aldrich chemicals, Mo). Following the removal of unbound secondary antibody by washing the immunoreactive bands were identified by chemiluminescence detection (HRP detection kit, Amersham Biosciences UK Ltd. Buckinghamshire, England) (Sloane et al, 2005). Antibody Binding Assays (MAP) To assess antibodies binding to a variety of common viruses, including human influenza viruses, globally different IVIG products were subjected to bead-based multianalyte profiling (MAP) at Rules-Based Medicine (www.rulesbasedmedicine.com) (Austin, Texas, USA), as described in detail by Bertenshaw and colleagues (Bertenshaw et al, 2008). The Influenza antigens used for the MAP analyses were from sonicates of purified Influenza A H3N2 strains A/Texas/1/1977 and A/Shandong/9/1993; and Influenza B, B/Hong Kong/5/1972. Viral microarray For binding of IVIG antibodies to influenza antigens, whole influenza antigen was robotically spotted in a protein array, and detected by influenza-specific antibodies in IVIG. Serial dilutions of influenza strains A/Sydney/5/1997 (H3N2) and A/Johannesburg/82/1996 (H1N1), Tetanus toxoid, and BSA, were spotted (10nl/spot) onto nitrocellulose coated glass slides (Schleicher & Schuell, Dassel, Germany) using a Piezorray robotic arrayer (Perkin Elmer). Slides were cured overnight at 4ºC then blocked with 5% (w/v) skimmed milk in PBS. One slide without milk blocking was stained with Ponceau Red to visualise total protein. Antibodies in IVIGs to human and avian H5N1 influenzas IVIGs are highly-purified IgG preparations prepared from large pools of human plasma and the immunoglobulin subtypes in IVIG formulations (i.e., IgG1, IgG2, etc) reflect those of normal plasma. The highly-purified nature of IVIGs is shown in Figure 1 of a protein stained 2D-gel electrophoresis map, and revealed heavy and light chains of these natural antibody preparations and their broad pI properties. In antibody binding studies using multiplex (Rules Based Medicine) and microarray binding assays (Figure 2A and B), there were quantitative and qualitative Figure 1. 2D gel electrophoretic profile of a representative IVIG preparation, with IPG isoelectric fractionation over 4-7 pI range. The separation of protein molecular weight standard markers in the acrylamide separation phase is shown on the left hand side of the 2D-separated immunoglobulin fractions. Note, the respective IVIG formulations used in this study (Table 1) have been deidentified, but each assigned a label A-G. The formulation used for 2D analysis was product D. differences in anti-influenza antibody binding to the antigens of H3N2 influenza A/Shandong/9/93, both between different IVIG preparations (data not shown), and within batches of a single IVIG preparation. Along with quantitative differences in antibody binding of immobilized, H1N1 and H3N2 strains of human influenza ( Figure 2B). The relative binding of human antibodies in IVIGs was compared by the multianalyte profiling of antibodies to antigens of a variety of other viruses that commonly infect humans, namely herpes simplex virus (HSV), human papilloma virus (HPV), Influenza B (Infl B), respiratory syncytial virus (RSV) and Varicella Zoster (V. Zoster). As might be predicted the antibody responses against the highly antigenic herpes simplex and influenza A viruses that commonly infect humans predominate over quantitative responses to the other viruses examined. Comparison of IVIG antibody binding to H5N1 and human influenzas For comparison with our studies of H5N1 influenza proteins separated by SDS-PAGE, shown in Figure 3 are the protein bandings of the most prominent proteins of 3 human influenza strains. Included are their relative molecular weights and understood per virion abundance (Privalsky and Penhoet, 1978;Lamb and Krug, 2001;Lynch et al, 2008). Between the protein profiles of the H1N1 and H3N2 strains, the stand out feature is the faster migrating HA (and oligomer) bands of H1N1 compared with H3N2 virus, reflecting glycosylation differences (namely, H3 > H1) (Lynch et al, 2008). For our studies we have equalized viral loading onto SDS-PAGE gels on the basis of protein equivalence, using the most abundant and detectable protein of influenza, M1, as an internal standard (Lynch et al, 2008 studies of human antibodies in Australian IVIG, that similar binding activities for avian H5N1 influenza proteins are widely found in all of the different IVIG preparations (Figure 4). IVIG antibodies from the various products bound equivalently to the highly conserved M1 protein across H5N1, H1N1 and H3N2 influenza virusus. From these immunoblotting comparisons we find equivalent IVIG antibody binding to the abundant and conserved internal M1 and NP proteins between H5N1, H1N1 and H3N2 isolates. With minimal binding to the far more heterogeneous HA and NA envelope proteins of H5N1 compared to the human isolates, for which the latter show marked binding activities to the surface and immunodominant HA proteins (including oligomeric forms) of each of the human strains ( Figure 4). When matched side-by-side differences in binding profiles for the human influenza isolates between the IVIG preparations undoubtedly reflects differences in the environmental strain-specific exposure and vaccination between the respective donor communities. As reported earlier (Lynch et al, 2008) and here, low levels of antibodies that bind to the HA of H5N1 are revealed by long exposure times. The specificity and heterosubtypic properties of this H5N1 antibody binding activity in IVIG and normal human serum was demonstrated by various means, including selective depletion by adsorption to human H3N2 and H1N1 influenza proteins in the absence of quantitative nonspecific depletion of the total IVIG pool of IgGs (Lynch et al, 2008;Stelzer-Braid et al, 2008;and unpublished data). Heterosubtypic human antibodies to avian influenza Optimization of chemiluminescence exposure times for avian H5N1 immunoblots has enabled more finite detection of human antibodies in IVIGs to many of the proteins of H5N1 (Lynch et al, 2008). This is further demonstrated in Figure 5, where several different avian isolates were examined by SDS-PAGE immunoblotting and compared directly with human H1N1 influenza at a 1/10 lower protein loading. As demonstrated in our earlier report, when optimized for signal detection human antibodies in globally different IVIG preparations revealed antibodies that bind with many of the H5N1 protein, none-the-least to H5 HA (Lynch et al, 2008). The faster migrating H5 HA on SDS-PAGE is indicative of lower total carbohydrate compositions than for the HA of H1 or H3 of human influenza viruses. These observations were made from consistent findings for the respective avian strains but we acknowledge a limited sampling and need for a more expansive examination of additional avian isolates. Consistent in our studies were apparent differences in the quantitative binding of antibodies in IVIGs for avian HAs with H9> H5> minimal if any H7 binding; and qualitative binding showing prominent apparent binding to the high molecular weight polymeric forms of H9 HA compared to H5 HA. More in-depth examination of avian strains is now needed to access and substantiate indicated differences in human IVIG antibody binding profiles between strains. Human antibodies in IVIG protect against H5N1 infection in cell culture As indicated from the comparative immunoblotting profiling of IVIGs antibodies above the low levels of heterosubtypic human antibodies that bind to H5N1 HA is of the order of 100-1000 times less than antibody binding to human influenza (H3N2, H1N1) HAs. Strain-specific human serum antibodies routinely give high neutralization titres in, in vitro, infection assays of human influenza infection. In contrast, any human antibody disruption of infection of a completely foreign influenza strain, such as H5N1, would likely be minimal, particularly considering the level of binding to H5N1 observed above. Despite this, when northern hemisphere IVIGs were tested in cell culture for H5N1 neutralization, we observed bona fide protection, albeit with modest titres with 100% inhibition (Table 3). These activities were similar to the neutralization levels observed within the same experiments in assays of southern hemisphere, Australian-derived IVIG, as reported earlier (Lynch et al, 2008) and duplicated her for direct comparison. Each of the IVIG preparations inhibited infections of 3 strains of H5N1 (i.e., 2 clade representative strains -one from human and two from avian infections), when tested in both Vero and MDCK cell culture infections. As each of the IVIG preparations inhibit the infections of 3 strains of H5N1 but not the infection of H7N7 virus, strain-specific heterosubtypic protection is implicated. The individual and combined roles of anti-HA and NA antibodies (as well as the more abundant antibodies to NP, M1, NS2) in providing this protection remain to be defined. DISCUSSION H5N1 continues to be a major global health concern, with 60% mortality among humans and 100% among some bird species. Understanding the immune mechanisms that underpin the immune protection in human survivors is therefore of fundamental importance, in order to firstly understand natural protection against any future expansion of H5N1 or other emerging new influenzas (e.g., the 2009 H1N1), and secondly to harness any heterosubtypic immune protection through the design of new vaccines and antivirals. Undoubtedly, innate and heterosubtypic cellular immune mechanisms protect against infection or disease of novel influenza viruses, but these are unlikely to explain the observed 40% level of 'natural' protection against H5N1 human infections. We believe our findings of heterosubtypic anti-H5N1 reactive antibodies in different unexposed populations offer an insight into a tenable process of adaptive antibody immunity to protect against novel influenzas such as H5N1. Following on from our earlier reports (Lynch et al, 2008;Stelzer-Braid et al, 2008), where we identified anti-H5N1 directed human antibodies in sera and IVIG preparations from Australian blood donors we now show that IVIGs from different geographic populations from the northern hemisphere, and notably from European blood donations collected prior to the emergence of avian H5N1 influenza in Europe, similarly share this property. This suggests that some preexisting heterosubtypic 'herd' antibody immunity to avian H5N1 influenza may be a universal characteristic of populations across the globe. To demarcate a heterosubtypic response from the well-studied 'seasonal' adaptive response that current vaccines stimulate we refer to this particular type of humoral immunity as a 'seasoned' adaptive response. While the value of strain-specific antibodies to protect against seasonal influenzas is well appreciated and understood, any suggestion of broad acting antibodies to protect against hitherto unseen novel influenza viruses (e.g., from birds or pigs) has till now been overlooked. From our observations we propose that pre-existing heterosubtypic antibodies in the community may offer a molecular mechanism to explain protection against H5N1 and past pandemic strains, and together with innate and heterosubtypic cellular immune protection could provide a significant barrier of protection for novel viruses. Our findings here and in previous reports give direct evidence that heterosubtypic 'seasoned' antibody responses to novel influenza viruses do exist, but what we do not yet know are the sites and mechanisms involved, and most importantly, in the wider context the significance of this response for natural protection against emerging infections and whether it could be harnessed for new styled anti-virals. This proposition gives rise to a number of predictions, which themselves can be examined and challenged, and if found to be of value, appropriately validated. These relate to the influenza protein epitopes that heterosubtypic antibodies target, and how an adaptive heterosubtypic humoral response may be established and progressed. We predict the development of antibodies to sites that are highly-conserved across influenza viruses, such that, when a new and antigenically novel influenza virus arises antibodies to exposed conserved sites, irrespective of the strain, could bind and disrupt viral infection and growth. In support of this concept the extent of binding of heterosubtypic antibodies of human serum and IVIGs to influenza target proteins directly correlates with the extent of the viral protein cross-strain sequence conservation. In making this comparison we benefit from the bioinformatic assessments of Bui et al (2007) and Heiny et al (2008) of conserved sites across a wide variety of influenza strains. Particularly significant is the study by Heiny and colleagues involving examination of 36,343 influenza A sequences of linear stretches of 9 amino acids or more that were conserved across influenza subtypes (Heiny et al, 2007). On those criteria, they identified a wide range of conserved sequences between the different proteins. Notably, they found that 50% of the protein sequence of polymerase protein PB2 is widely conserved followed by 36% conservation of PB1, 27% of M1, 25% of NP, 16% of PA, and only 2% of HA. On this basis, none of the proteins NA, PB1-F2, M2, NS1 or NS2 had fully conserved regions across this large sequence data base (Heiny et al, 2007). Within the sensitivity constraints of our studies and restricted to the most abundant proteins (see Figure 3) we broadly found that antibody binding was preferential for the H5N1 proteins with the highest sequence conservations with human influenzas ( Figure 5). One notable example was that of the most abundant influenza protein, the matrix protein M1 of H5N1 with 27% conserved stretches of sequence, which we found to have equivalent IVIG antibody binding as for the M1 protein of human H1N1 and H3N2 influenza strains (also see Lynch et al, 2008). These considerations do not however take into account the contributions of other heterosubtypic antibody responses to conserved, linear regions of less than 9 amino acids, or to conserved discontinuous epitopes. Unequivocal validation of each of the antibody-epitope interactions is now required to define the respective antibody binding kinetics and saturation, and by peptide and protease fragmentation studies coupled with mass spectrometry to definitively resolve the targetepitopes and match against conserved sites predicted from bioinformatics data-base screening and molecular modelling. In this study we used IVIGs as a tool to broadly study human antibodies for H5N1. However, IVIGs as a potential therapeutic option in the event of the sudden emergence of a HPI pandemic could also be considered. IVIGs are well-tolerated therapies for primary and secondary immune-deficiencies and autoimmune disorders (Simon and Spath, 2003). However, while IVIGs and hyper-immune immunoglobulins are highly valued and studied therapies for a variety of infectious agents, including CMV and hepatitis viruses (Gupta et al, 1996), there is a notable paucity of information in the literature on the anti-viral properties of IVIGs for common viral agents, such as for influenzas. If it was possible to identify a cohort of individuals with solid pre-existing seasoned immunity for hyper-immune heterosubtypic anti-influenza antibodies for passive immune therapy, this could add an additional therapeutic asset to complement small molecule anti-virals and vaccines. If hyper-immune anti-influenza IVIG were possible, in addition to harvesting pandemic convalescent influenza antibodies they would offer an already licensed production-capable fallback therapy option during a HPI pandemic. The passive transfer of antibodies for the protection of infection-naïve individuals at risk from pandemic influenza was successfully deployed in 1918 and more recently in the successful treatment of an individual infected with H5N1, and to this day convalescent 'Spanish Flu' survivors have retained protective antibody responses to the 1918 H1N1 influenza (Luke et al, 2006). The antibody repertoires of convalescent individuals would be particularly beneficial, as they would possess a complement of both seasoned and seasonal antibodies to protect against epidemic and pandemic influenzas. Furthermore, as there is growing evidence implicating the prominent role of secondary infections in the mortality of pandemic influenzas, IVIGs, as broad anti-infective agents, may also be beneficial in a pandemic as a backup for antibiotics, to keep secondary infections in check (e.g., bacterial pneumonia, CMV, tuberculosis) (Klugman et al, 2009a;Klugman et al, 2009b) and to suppress overt host cytokine-storm responses particularly in younger individuals, who are prime targets for pandemic influenza mortality. As the quest in the design of universal vaccines and inhibitors for influenza gains momentum we may well learn many essential lessons from how heterosubtypic antibody immunity has developed in nature to protect against new infectious challenges. ACKNOWLEDGMENTS Competitive Grant Support for the study of IVIG and avian H5N1 antibody immunity was gained from the Australian Red Cross Blood Service Grant for GWL and from the University of Sydney for GWL and JSS. We wish to acknowledge the Ramaciotti Centre for Gene Function Analysis (The University of New South Wales, Randwick, NSW Australia) for access and use of the Piezorrayer.	2014-10-01T00:00:00.000Z	2009-12-01T00:00:00.000	{ "year": 2009, "sha1": "fb82dd60b300c0d07c0bf2732e20eda467214fd7", "oa_license": "CCBYNC", "oa_url": "https://www.omicsonline.org/open-access/heterosubtypic-antiavian-hn-influenza-antibodies-in-intravenous-immunoglobulins-from-globally-separate-populations-protect-against-hn-infection-in-cell-culture-1747-0862.1000038.pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "fb82dd60b300c0d07c0bf2732e20eda467214fd7", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
248451312	pes2o/s2orc	v3-fos-license	Promotion of seed germination and early plant growth by KNO3 and light spectra in Ocimum tenuiflorum using a plant factory The plant factory with artificial light (PFAL) is a novel cultivation system of agriculture technology for crop production under controlled-environment conditions. However, there are a number of issues relating to low quality of seed germination and seedling vigor that lead to decreased crop yields. The present study investigates the optimal KNO3 concentration for seed germination, and the influence of different light spectra on early plant growth in holy basil (Ocimum tenuiflorum) under a PFAL system. Experiment 1 investigated the effects of KNO3 concentration (0, 0.2, 0.4 and 0.6%) on germination of seeds primed for 24 h under white Light emitting diodes (LED). Results show that sowing holy basil seeds in 0.4% KNO3 enhanced seed germination percentage (GP) and germination index (GI), while decreasing mean germination time (MGT). Experiment 2 investigated the effect of four light spectra on seed germination and early plant growth by sowing with 0 and 0.4% KNO3 and germinating for 15 days continuously under different monochromatic light settings: white, red, green and blue in PFAL. It was found that the green spectrum positively affected shoot and root length, and also decreased shortened MGT at 0 and 0.4% KNO3 when compared with other light treatments. Additionally, pre-cultivated seedlings under the green spectrum showed significant improvement in the early plant growth for all holy basil varieties at 15 days after transplanting by promoting stem length, stem diameter, plant width, fresh weights of shoot and root, and dry weights of shoot and root. These findings could be useful in developing seed priming and light treatments to enhance seed germination and seedling quality of holy basil resulting in increased crop production under PFAL. colds, and for influenza [11][12][13][14] . Holy basil is an important commercial species with estimated annual trade of 2,000 to 5,000 metric tons in India 15 . In 2020, holy basil was used to as an adaptogens protect the body from physical and mental stress which is estimated to expand not less than 1.5 billion USD according to transparencymarketresearch.com. However, the agricultural industry faces a number of issues related to the low quality of seed germination and seedling vigor in holy basil, which leads to decrease in crop yields. Low seed germination has been shown to relate to the mucilaginous layer (seed gum) 16 . Abraham 16 reported that the mucilage is apectinous matrix, consisting of considerable amounts of unesterified galacturonic acid with a large capacity for hydration that act as a reservoir blocking the absorption of oxygen while the seeds absorb water [17][18][19] . Moreover, the high accumulation in some phenolic and polyphenol compounds (benzoic acid and p-hydroxybenzoic acid) might exert potentially negative effects on seed germination and seedling development by decreasing enzyme activity in cellular functions and impairing the progression of cell division 20,21 . There are many ways to stimulate the germination of mucus-covered seeds, such as washing seed gum with ascorbic boric and HgCl 2 22,23 or soaking the seeds in a solution that promotes germination, such as GA 3 KNO 3 kinetin and ethrel 24 . Seed with mucous or hard seed coats can be improved by acid soaking before planting 20 . However, this method causes high seed mortality when used with large quantities of seeds 25 . Seed priming involves preparing seeds before planting by soaking in water or chemical solution at the appropriate temperature and time-period. This is done to stimulate the early events of germination and various biochemical or metabolic processes, before drying the seed back to its original moisture condition 26 . It also causes reorganization of membrane and repairs damaged cells and organs 27 . Priming increase the uniformity, germination speed and the growth of seedling 28 . Seed priming includes hydropriming, hormonal priming, osmopriming, matrix priming. Hydropriming simply uses clean water and is therefore safe for the user, reduces costs, and does not leave toxic or chemical residues in either seeds nor the environment 29 . Osmopriming involves soaking of seeds in low water potential solutions and reduces the rate of water imbibition 27 . Osmotica used for seed priming include organic salts, such as polyethylene glycol (PEG), manitol and sorbitol, and are most commonly used to adjust osmotic potential. Inorganic salts, such as KNO 3 , CaCl 2 , KH 2 PO 4 30 increase nitrogen and other nutrients needed for protein synthesis while the seeds germinate. KNO 3 is most widely used in seed priming to improve seed germination due to it is an ionic salt of potassium ions (K + ) and nitrate ions (NO 3 − ). It occurs as an alkali metal nitrate and is a major essential plant nutrient. Nitrogen is a component of many biomolecules in plant cells and helps seeds synthesize proteins, which has an impact on seed quality 31 . Additionally, K + dissolves in the cytoplasm and vacuole and acts to maintain the osmotic potential. K + is also associated in the stimulation of over 40 type of enzyme, especially enzymes in photosynthesis and respiration, as well as those used in the synthesis of starch and protein which help maintain the firmness of plant cells [32][33][34][35] . The duration of seed priming is critical and has been reported for many crop plants 28 . Seeds of pepper primed in PEG for six days produced more abnormal seedlings than seeds primed for four or five days 36 . In muskmelon studies, seeds are primed from 16 h 37 to ten days 38 . Hydropriming at 12 h improve seedling vigor and germination percentage 39 , in addition it enhances seedling shoot and stem vigor index 40 of basil. One of the most important abiotic parameters for plant growth throughout the life cycle is light availability 41 . Solar light consists of electromagnetic energy with wavelengths ranging from 400 to 700 nm (violet, blue, green, yellow, orange and red). Plants have evolved light absorbing molecules that enable organisms to respond to changes in the light quality that affects various physiological processes depending on the species, developmental stage, or studied organ 42 . Furthermore, light acts as an environmental signal controlling the plant photomorphogenetic responses, including the transition from one development stage to the next 43 . Green light stimulates seed germination via the early elongation of the stems by antagonizing growth inhibition processes in Arabidopsis 44 . Blue light has a significant impact on seed germination by decreasing the germination percentage and mean germination time of Brassica napus 45 . In recent years, the utilization of plant factories with artificial lighting (PFAL) has become more commonplace for plant production purposes, and is considered an alternative crop plant production method in response to climate change 46,47 . PFAL is a modern agricultural system using advanced technologies for plant cultivation in a fully controlled environment, including: light, humidity, carbon dioxide, temperature, water and nutrient 48 . Light is one of the most important environmental factors affecting the plant growth, plant development and crop yields, especially in plant factory or indoor farming settings. In order to maximize the biomass productivity in a PFAL, it is critical to control the optimal light quantity (intensity and duration) and quality (wavelength composition) to improve the quantity and quality of crop plants. Currently, much research on the utilization of PFAL has been conducted on controlling seedlings in terms of stem diameter, hypocotyl length, fresh and dry weight, compactness and root development. Each is important to improve the success rate of plant production by manipulating the growth conditions [49][50][51] . Moreover, holy basil is commercially cultivated through transplanting of seedlings, and often the farmer and commercial growers suffer from major losses due to low quality seeds. It is therefore essential to assess the germination potential and vigor in order to ensure an optimum crop stand and herb yield. Methods for improving sweet basil seed quality has been reported in various studies, however, there is no method to test for germination of holy basil 52 . Accordingly, this study aims to evaluate the optimal concentration of KNO 3 for seed germination, and the influence of different monochromatic light spectra on seedling vigor and early plant growth in holy basil under a PFAL system. This study represents the first study reporting a priming effect of KNO 3 concentration on seed germination and the influence of different monochromatic light on early plant growth of holy basil. Moreover, it contributes towards the improvement of seed technology utilizing LED technology. Materials and methods Seed source. Four varieties of holy basil, including two types of green and two types of red, were obtained from three commercial seed companies in Thailand as shown in Table 1 from each variety and were primed with KNO 3 solution at four different concentrations: 0%, (KNO3 0 g L −1 ) 0.2% (KNO 3 2.0 g L -1 ), 0.4% (KNO 3 4.0 g L -1 ) and 0.6% (KNO 3 6.0 g L -1 ) for 24 h at 25 ± 1 °C. The seeds were placed on Petri dishes with KNO 3 solution, with petri dish lids then placed on top. Afterwards, seeds were rinsed thoroughly with reverse osmosis (RO) water and dried back to original moisture contents at room temperature for 2 days. Primed seeds with 0% KNO 3 (hydropriming) were maintained as control for comparison. Experimental site and germination test. After priming, seed germination tests were conducted. Four replicates of 50 seeds each were used for each treatment, placed on top of the filter paper with 10 mL of RO water in a Petri dish. Seeds of each treatment were set in each of the five shelves available in the PFAL. Seed germination tests were conducted in an environmentally-controlled room under PFAL system of National Center for Genetic Engineering and Biotechnology (BIOTEC), National Science and Technology Development Agency (NSTDA), Pathum Thani, Thailand. The environmental conditions consisted of 150 µmol m −2 s −1 of white LED lights for PPFD with 12 h d −1 photoperiod and 25 ± 1 °C temperature. Germinated seeds featuring cotyledons roots were counted every three days for 15 days. Then, the germination percentage (GP) on day 15 after sowing was calculated. The germination index (GI) and mean germination time (MGT) were calculated as GI = ∑(G t /T t ) and MGT = ∑(G t × T t )/∑G t , respectively 53,54 , where G t is the number of germinated seeds on Day t, T t is time corresponding to G t in days. The results of KNO 3 concentration from this experiment were taken into consideration for selecting the appropriate treatment to be used in the second experiment. Statistical analysis. The experiment was arranged as a completely randomized design (CRD) with four replications and 50 seeds per replicate for each concentration. The priming concentration were 0%, 0.2%, 0.4% and 0.6% KNO 3 . GP, MGT and GI were subjected to analysis of variance (ANOVA). The differences between the means were compared using Duncan Multiple Range's test (P < 0.05). Experiment 2: light spectrum on seed germination and seedling characteristics. Monochromatic light treatment and Statistical analysis. According to the data form the first experiment, 0.4% KNO 3 was chosen for seed priming in the PFAL system. Holy basil seeds were primed with 0% and 0.4% KNO 3 for 24 h at 25 ± 1 °C. The seeds were rinsed with RO water and dried back to their original moisture contents (less than 15% moisture content) at room temperature for 2 days. Four biological replicates were performed for each sampling concentration, each one containing 50 seeds. The seeds from each variety were sown on a sponge mat with a 50-cell germination (cell size 24 × 11.5 × 3 cm) (ESPEC Corp., Japan), and placed on foam trays ( Fig. 1A) containing RO water with water saturation content at 100% moisture content. For seed germination, a total 200 seeds of each variety from the trays were moved to the shelves under controlled environmental conditions. To maintain a high moisture content, the germination trays were covered with clear plastic sheets and watered daily with RO water. The environmental conditions consisted of 150 µmol m −2 s −1 of white LEDs, monochromatic blue lights (λ = 400-500 nm, peak at 450 nm), green (λ = 500-600 nm, peak at 525 nm) and red (λ = 600-700 nm, peak at 660 nm) ( Supplementary Fig. S1) of PPFD with 12 h d −1 photoperiod, 70 ± 5% relative humidity (RH), 1000 ± 300 µmol mol −1 (ppm) CO 2 concentration and 25 ± 1 °C temperature. The number of germinated seeds was recorded daily over a period of 15 days. A seed was considered as germinated when the hypocotyl and cotyledons protruded from the seed coat. At the end of this period, the GP (%), MGT and GI were calculated. Root and shoot lengths were determined by digital photographs using ImageJ v 1.5.3e software 55 . The same randomly selected seedlings were collected and transferred to the hydroponics system to evaluate the seedling performance attributes after pre-treatment with different monochromatic light for 15 days. The experiment was arranged as a completely randomized design (CRD) with 8 treatments (0% and 0.4% KNO 3 combined with LEDs light that is white, red, green and blue light) each with four replications and each replicate of 50 seeds. Germination percentage, mean germination time and germination index were subjected to analysis of variance (ANOVA). The differences between the means were compared using Duncan Multiple Range's test (P < 0.05). Seedling performance and statistical analysis. Based on the monochromatic light treatments, holy basil seedlings at 15 days after sowing were transplanted into plant production room in the PFAL system which 200 µmol m 2 s −1 white light for PPFD with 12 h d −1 photoperiod, 70 ± 5% relative humidity (RH), 1000 ± 300 µmol mol −1 (ppm) CO 2 concentration and 25 ± 1 °C temperature. Seedlings featuring expanded cotyledons, elongated and www.nature.com/scientificreports/ hypocotyl and also enhanced root length were transferred to hydroponic foam boards ( Fig 1B). Foam boards with seedlings were put into the cultivation bed ( Fig 1C) and allowed to grow in a hydroponics system utilizing DFT (deep flow water technique). All seedlings in each treatment were grown in a dedicated compartment at the same electrical conductivity (EC) of the nutrient solution (0.8 mS cm −1 ). The selected plants were provided with a modified Enshi solution consisting of 220 g L −1 Ca(NO 3 ) 2 ·4H 2 O, 120 g L −1 KNO3, 120 g L −1 MgSO 4 ·7H 2 O, 60 g L −1 KH2PO4, 10 g L −1 Fe-EDDHA, 10 g L −1 Fe-DTPA, 10 g L −1 NIC-SPRAY, 2 g L −1 Mn-EDTA, 100 mg L −1 NiSO 4 6H 2 O, 100 mg L −1 NaMoO 4 ·H 2 O. After 15 days of transplanting, plant growth parameters; stem length (cm), plant width (cm) and stem diameter (mm) were measured with a ruler and digital caliper. Shoot and root fresh weights were collected. Then, dry weight was obtained by drying in an oven set at 50 °C for 3 days, followed by weighing to determine average seedling dry weight per replication. The experiment was arranged as a completely randomized design (CRD) with four replications and fifteen seedlings per replicate for each treatment. Four LED treatments primed with 0% and 0.4% KNO 3 were subjected to analysis of variance (ANOVA). The differences between the means were compared using Duncan Multiple Range's test (P < 0.05). Research involving plants. All the relevant institutional, national, and international guidelines and legislation have been followed. Table 2. The results of the current study indicate that seed priming with KNO 3 improved the seed establishment of holy basil on the top of paper under 150 µmol m −2 s −1 of white LEDs at 25 °C. Holy basil seeds primed with 0.4% KNO 3 had maximum GP and GI among three varieties, including G-NT (72.5%), R-JT (90%) and R-TS (88.0%). For MGT, there was no significant differences between the four concentrations of KNO 3 . Moreover, the patterns of seed germination of holy basil varieties for 15 days after sowing indicated the stimulatory impact of KNO 3 (Fig. 2). The GP at 0.4% KNO 3 showed a similar increasing pattern after six days of sowing among all four holy basil varieties. Thus, we focused our investigation on the effects of 0.4% KNO 3 with monochromatic LEDs under fully controlled environment in plant factory system. at 0% (i.e. as control) and 0.4% were used for priming holy basil seeds for 24 h which were treated with four monochromatic LEDs under fully controlled conditions. Seedling establishment including GP, MGT and GI at 15 days after sowing (before transplanting) is shown in Table 3, the GP of the four holy basil varieties primed with 0% and 0.4% of KNO 3 were not significantly different among light treatments (except R-JT at 0.4% KNO 3 ). There were no significant differences observed in the GI parameter of the four light treatments across all varieties. Seeds primed with 0% and 4% KNO 3 under monochromatic blue light (λ = 450 nm) and monochromatic green light (λ = 523 nm) resulted in a lower MGT than white LEDs and monochromatic red light (λ = 660 nm), while maximum MGT was observed with white LEDs. Although all seed varieties treated with monochromatic blue light germinated earlier, shoot lengths were shorter than for green light. Only monochromatic green light showed significant improvement in shoot length for all four varieties in both priming treatments at 0% and 0.4% KNO 3 at 15 days after sowing (Fig. 3A,B), However, light treatments did not exert any significant effect on root length (Fig. 3C,D). The phenotypes of holy basil seedlings under four light treatments at 15 days after sowing (before transplanting in PFAL system) are shown in Supplementary Fig. S2. The results indicate that monochromatic green light can decrease germination time and promote shoot length growth before transplanting among all four holy basil varieties. Results Early plant growth under PFAL system. Based on the above results, all holy basil seedlings from each monochromatic light treatment with 0% and 0.4% KNO 3 priming were transplanted under fully controlled conditions in the plant factory to determine the effects on early plant growth. Morphology of the plants 15 days after transplanting is displayed in Fig. 4. Morphological evaluation of seedling quality, including stem length, plant width and stem diameter of four varieties at 15 days after transplanting are presented in Fig. 4. Data revealed that seedlings under green light treatment primed with both KNO 3 concentrations featured significantly greater stem length and plant width when compared to those from white, red and blue light treatments (Fig. 5A-D). Seedlings from three monochromic lights with 0% KNO 3 priming showed no significant differences for the stem diameter in G-NT and G-TS, however, it was higher than white LEDs (Fig. 5E). Nonetheless, seedlings of all varieties under green light treatment with 0.4% KNO 3 priming had a significant greater or broader stem diameter compared to light treatments (Fig. 5F). Green light treatment also showed greater fresh and dry weight accumulation of shoot after growing with white LEDs for 15 days (Fig. 6A,B). Seedlings with pre-treatment for 15 days with monochromatic green light also induced greater root fresh weight in all varieties when compared to other light treatments at 15 days after transplanting; however, there were significant differences in root dry weight among light treatments with 0.4% KNO 3 priming, except for both red holy basils. (Supplementary Table S1). In general, our study indicates that monochromic green light might be the best available method owing to higher complete germination viability and seedling establishment leading to plant vigor and plant development. Table 2. Final seed germination percentage (GP) and mean emergence time (MGT) and germination index (GI) at 15 days after sowing of four holy basils as affected by seed priming with KNO 3 , on top of the filter paper with white LED. Data are mean values ± SE (n = 4) with four fifty seeds in a replication. Different letters in the same column indicate significant difference at p < 0.05. * indicates significant difference. "ns" indicates no significant difference. Discussion Seed priming improved both seedling growth and reduction of seedling germination time leading to higher grain yields 56,57 . Priming enhances the metabolic processes during early phase of germination before radicle protrusion and improves seed respiration processes 58,59 . Moreover, seed priming with KNO 3 has been shown to influence final seed germination percentages by decreasing the germination times, seed vigor and uniformity of germination in various plant species [60][61][62][63][64][65] . Similarly in our study, we found the difference in germination performances between KNO 3 concentrations. Results from experiment 1 showed that KNO 3 priming influenced seed germination percentage, mean germination time and germination index among four holy basil varieties. Moreover, the results shown in Table 2 indicate that priming of holy basil seeds with 0.4% KNO 3 caused greater final GP and GI than other concentrations. Several studies have found that higher concentrations of KNO 3 for seed priming affect seed germination by inhibiting GP and MGT in many plants [65][66][67][68] . A similar trend in slight suppression of seed germination pattern in this study was observed for high concentrations of KNO 3 (0.6%) across all holy basil varieties. The findings of the present study indicate that the performance of all holy basil varieties primed with 0.4% KNO 3 was appreciably improved when subsequently germinated on top of the filter paper with white LEDs in plant factory. The pattern of seed germination and germination index were almost the same for the four varieties ( Table 2), meaning that seed priming with 0.4% KNO 3 was useful in terms of GP and GI for all holy basil varieties. In contrast, seed priming with 0% and 0.4% of KNO 3 did not significantly influence GP and GI according to light treatments under fully controlled environment in PFAL using the sponge mat (Table 3). There are some reports explaining the performance of seed germination according to many factors including types of substrate, growth media, and also environmental factors such as light, oxygen, water, temperature and plant species 69,70 . A good substrate provides adequate high water retention capacity necessary for seed germination of holy basil. Since the water absorption rate and biochemical reactions are also key factors to stimulate the seed germination, the increase in seed germination percentage could be a reflection of the high moisture content in agricultural substrate 71,72 . Recently, many research studies on the utilization of LED technology on plant growth, development, and morphology have been carried out under in vitro conditions and as indoor experiments 73,74 . Control of seedling growth and quality traits such as hypocotyl length, root length and stem diameter is very important in promoting www.nature.com/scientificreports/ success rate of plant production under manipulated environmental conditions 50 . However, the influence of light spectra on seed germination and early plant growth under PFAL systems has been rarely reported. In the present study, promotion of both seed germination and seedling quality under monochromatic lights is reported for the first time in holy basil. The results of this study demonstrate that monochromatic green light results in decreased MGT and significantly promotes shoot length when compared with other light treatments of holy basil primed with 0.4% KNO 3 (at 15 days after sowing). The results are consistent across the four holy basil varieties and have the same trends for MGT and shoot length ( Fig. 3 and Table 3). These results are also in agreement with observations of Pierik 75 and Zhang 76 who reported that seedlings grown under green light also commonly show typical shade avoidance responses such as increased shoot growth and plant height. Moreover, green light is efficiently absorbed in inner canopy levels during plant photosynthesis (Folta and Maruhnich 77 ) and might stimulate growth through phytochrome and cryptochrome activity leading to morphogenetic change 78 Under early plant growth after 15 days of transplanting, the seedling from green light pre-treatment caused the greatest effects on stem length, plant width, and stem diameter in holy basil seedling (Fig. 5). This would likely Table 3. Final seed germination percentage (GP) and mean emergence time (MGT) and germination index (GI) before transplanting at 15 days after sowing of four holy basil varieties by seed priming with KNO 3 , under different LED light treatments; white (W), monochromic red (R), green (G) and blue (B) in the PFAL system. Data are mean values ± SE (n = 4) with four fifty seeds in a replication. Different letters in the same column indicate significant difference at p < 0.05. * indicates significant difference. "ns" indicates no significant difference. KNO 3 concentration Light treatment Seed germination (%) www.nature.com/scientificreports/ greater more leaf expansion and shoot length, and thereby cause plants to accumulate increased total carbon gain and lead to faster growth and biomass 82 . Interestingly, our result showed the same increased tendency in fresh and dry weights of shoot for all holy basil varieties (Fig. 6). Thus, seed germination, and seedling vigor are the two essential trait parameters determining the success in crop production 83 . These results suggest that pretreatment under monochromatic green light had a significant effect on seedling quality before transplanting as well as for early plant growth that may positively affect plant vegetative growth. In addition, the monochromatic green light triggered specific responses in the growth and morphology of holy basil seedlings that were demonstrable even after the seedling transplanting to PFAL conditions. Still, further research is needed to understand the influence of monochromatic light spectrum on pigments, secondary metabolites and antioxidant systems during seed germination, which may also contribute to greater understanding in enhancing the productivity of many other economically important crops. Conclusions The results demonstrate an effective seed priming (with KNO 3 ) and pre-treatment with light spectrum approach for promoting seed germination and seedling performance attributes among four holy basil varieties sourced from three seed companies. Seed priming performed at several concentrations of KNO 3 found the highest seed germination rate at 0.4% of KNO 3 while placed on top of the filter paper with white LEDs. Under plant production processing in PFAL conditions, GP and GI of all holy basils did not differ between monochromatic light www.nature.com/scientificreports/ treatments. However, green light spectrum showed high seedling improvement before transplanting by reducing MGT and increasing shoot length. Furthermore, results after transplanting confirm the advantage of using green light spectrum to increase the production of high quality seedlings in holy basil. This should be a successful approach for improving the speed of seed germination, seedling establishment and seedling vigor, thus enhancing the productivity of holy basil under PFAL system. www.nature.com/scientificreports/	2022-05-01T06:23:17.834Z	2022-04-29T00:00:00.000	{ "year": 2022, "sha1": "e23f36e3069b2749b5745a266d8c4e46301f71ad", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41598-022-11001-5.pdf", "oa_status": "GOLD", "pdf_src": "SpringerNature", "pdf_hash": "6c2ab94714d693ff44f8c2920bc8bd02012acbca", "s2fieldsofstudy": [ "Agricultural And Food Sciences" ], "extfieldsofstudy": [ "Medicine" ] }
214612176	pes2o/s2orc	v3-fos-license	Azimuthal and polar anchoring energies of aligning layers structured by nonlinear laser lithography In spite of the fact that there are different techniques in the creation of the high-quality liquid crystals (LCs) alignment by means of various surfaces, the azimuthal and polar anchoring energies as well as the pre-tilt angle are important parameters to all of them. Here, the modified by a certain manner aligning layers, previously formed by nonlinear laser lithography (NLL), having high-quality nano-periodic grooves on Ti surfaces, recently proposed for LC alignment was studied. The change of the scanning speed of NLL in the process of nano-structured Ti surfaces and their further modification by means of ITO-coating, and deposition of polyimide film has enabled different aligning layers, whose main characteristics, namely azimuthal and polar anchoring energies, were measured. For the modified aligning layers, the dependencies of the twist and pre-tilt angles for LC cells filled by nematic E7 ({\Delta}{\epsilon}>0) and MLC-6609 ({\Delta}{\epsilon}<0) were obtained. Also the contact angle for droplets of isotropic liquid (glycerol), and nematic LCs was measured for the various values of the scanning speed during the laser processing. Introduction It is relevant to note that today's industrial production of the liquid crystal displays (LCDs) is very intense. The alignment of LCs is an important stage of the production of various LC devices. It is known that the most commonly encountered method of the the nano-structured aligning layer. One of the most important characteristics of the aligning layer certainly is AE, namely azimuthal anchoring energy (AAE W φ ) and polar anchoring energy (PAE W θ ) which will be described in this manuscript. By knowing the AE of aligning surfaces and having a way for controllable changing it, we can use these surfaces for different applications. For example, surfaces possessing strong AAE W φ is needed in order to receive the strong homogeneous alignment of LCs and can be used for storage some information. In contrary, to obtain the better characteristics in dynamic processes, namely the low-voltage switching of LCs from planar (or tilted) to homeotropic alignment (vertical, VA) or vice versa from planar to VA will take place for surfaces possessing weak AAE W φ and strong PAE W θ . It is well known that there are several calculation methods for AE values of aligning surfaces [22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37]. It is obvious that the orientation of the LCs depends on the AE between molecules of LC and aligning layer. Furthermore, any surfaces, including the aligning layers, are also characterised by surface tension γ S [38]. Consequently, some surfaces possess a hydrophobic property while other surfaces are characterized by a hydrophilic feature. Owing to this fact there might be different contributions of LC molecules orientation on the aligning surface in bulk of LC cell ( Figure 1) [39]. According to empirical Creagh-Kmetz's rule [38,39] the homeotropic alignment of nematic LC is obtained at γ S < γ LC while the planar alignment occurs at γ LC < γ S (γ S is the surface tension of the substrate and γ LC is the surface tension of LC). Recently it was shown that on the one hand, there has been certain exposure dependence of the contact angle of the nematic droplet E7 at the photosensitive PVCN-F surface [40] and on the other hand, the irradiation time dependence of the AAE W φ of PVCN-F surfaces was found [36]. It is important to note that the influence of the laser power on both the adhesion, surface free energy graphene treated by nonlinear laser lithography (NLL) [41], and contact angle of the water droplets deposited on the various treated graphene surfaces was studied [42]. Recently, we studied the nano-structured Ti layer by means of the NLL as aligning layer for the nematic LCs orientation [19,20]. For pure nano-structured Ti layers and Ti layers coated by polyimide ODAPI film the AAE W φ was determined by using the method of combined twist LC cell. It was shown that the value of the AAE W φ of aligning layer depends on various parameters of the NLL, namely laser pulse fluence (LPF, J) and scanning speed υ [19,20]. It was therefore logical the next stage study for characterisation of nano-structured Ti layer as an aligning surface of LCs in terms of PAE W θ , depending on various parameters of the laser processing. However, currently, there exist a few well-known methods based on the usage of the alternating electric field [22][23][24][25][26][27][28][29][30][31][32][33][34] to estimate the PAE W θ of the aligning layer. Therefore, LC cells besides of the aligning layer, additionally posses an optically-transparent current-conductive layer (for instance, ITO layer on glass substrate). In the case of the usage of Ti layer, which is also electrically conductive, after the laser processing, the nano-structured Ti layers are characterized by the periodic change of the nano-areas possessing both a non opticallytransparent current-conductive layer Ti and an optically-transparent electrically nonconductive pure glass. Unfortunately, because of that periodic changes the nanostructured Ti layer is not appropriate for measured of the PAE W θ , because the unbroken conductive layer with periodic grooves should be used as recently it was realized for nano-structured ITO layer with periodic grooves [21]. On this basis of above stated, in this manuscript modified nano-structured Ti layer additionally coated with ITO layer was used and studied. In this case the unbroken nano-periodic conductive ITO layers, by reiterating the nano-profile of the processed Ti layer, are observed. A new modified aligning layer was obtained as a result. As the main parameters of the aligning layers, the AAE W φ and the PAE W θ of pure ITO layer, already deposited on nano-structured Ti layer, and ITO layer additionally coated with polyimide PI2555 film were studied in this manuscript. Obviously, the different surfaces, treated by NLL with the use of various scanning speeds, and further modification of Ti layer by coating ITO and polymer layer, possess various quantities of AE, pretilt angle and contact angle of droplet of isotropic liquid and nematics. Materials To study the alignment properties of the modified nano-structured Ti layers, two nematic liquid crystals, namely E7 and MLC-6609, obtained by Licrystal, Merck (Darmstadt, Germany) was chosen. To obtain the planar alignment in the azimuthal plane of the both nematics, nmethyl-2-pyrrolidone solution of the polyimide PI2555 (HD MicroSystems, USA) in proportion 10:1 was used. The refractive index of the polyimide of PI2555 is n = 1.7 [49]. To study of the contact angle at the aligning layers the analytical reagent grade glycerol (C 3 H 8 O 3 ) with density ρ = 1.256 g/cm 3 at T = 20 o C and refraction index n D 20 = 1.4728 was used. The creation of modified nano-structured Ti layers The creation of Ti layers was as follows. At the beginning 300 nm thick Ti layer was deposited on a glass substrate, as were recently described [20]. To create the 5×5 mm 2 area of nano-structured Ti layers we used the experimental scheme of the NLL [50]. Ti layers were treated by means of the NLL with various scanning speed υ at the constant LPF J = 0.55 J/cm 2 , resulting in various periodic nano-structured Ti layers, possessing the certain period Λ and depth A. The next stage was a modification of the nano-structured Ti layers by coating of ~ 50 nm thick ITO layer on top of them. Such modified films were investigated as the first type of aligning layers (FTAL), schematically shown in Figure 2. with area 5×5 mm 2 and additionally coated with ITO layer (FTAL). Polymer film preparation About The determination of AAE W φ To determine AAE W φ the well-known method of combined twist LC cell [35,36], allowing to measure the twist angle φ, was used. For this aim combined twist LC cells, consisting of tested substrates, coated with modified films of both types, were made as recently described in [20]. The reference substrate was coated with the polyimide PI2555 processed with the rubbing technique [19,20]. An experimental scheme to measure of twist angle φ of the combined twist LC cells is detailed elsewhere [20,35,36]. The measurement of twist angle φ of the combined LC cell allows us to calculate the value AAE W φ of the aligning layer of tested substrate. The twist angle φ is related to the AAE W φ as follows [36][37][38]: where d is the thickness of the LC cell, φ 0 ~ 36 o is the angle between the easy axes of the reference and tested substrates; φ is the measured twist angle. The measurement of LC pretilt angle θ p Before the determination of PAE W θ the LC pretilt angle θ p measurements were carried out. The well known crystal rotation technique was used for this purpose and which is detailed elsewhere [51][52][53][54]. This method is to measure the optical transmission of LC cell as a function of the incident light angle. Here, LC cell was set between a polarizer and an analyzer, whose transmission axes are perpendicular to each other and make an angle of 45 degrees with the LC alignment direction. Therefore, the LC cell rotates around the perpendicular to the alignment direction axis. Consequently, the transmitted beam is measured for different angles of LC cell rotation. In this manuscript, the measurements of LC pretilt angle θp were carried out for LC cells, assembled with two the same substrates having either FTAL or STAL and filled by the nematic E7 (Δε > 0) or MLC-6609 (Δε < 0). Electric field method is well known as RV technique [29,30,33], which requires the measurement of phase retardation from voltage in contrast to classic method, knowing as Yokoyama-van Sprang technique [27] that requires the simultaneous measurement of the capacitance and optical phase. RV technique has a linear dependence, when the C is practically constant in some range of voltage (V min , V max ). By knowing phase retardation R 0 at initial voltage (V = 0), it was obtained the formula, which allows determination of the PAE W θ from a simple linear fit, as follows [29,30]: , α is a coefficient in range (2/π, 1), 0 J is a constant, K 1 and K 3 is the splay and bend elastic constant of nematic LC, d is the thickness of LC cell, θ p is the pretilt angle, \|\| ε and  ε represent parallel and perpendicular components of electric permittivity of LC. The phase retardation was measured by means of simple Senarmont compensator [55] consisting of a polarizer P, sample -LC cell, λ/4quarter-wave plate (QWP) and an analyzer -A. In this scheme two Glan-Thompson prisms were used as a polarizer P and an analizer A. Each optical element of Senarmont compensator was characterized by a definite azimuth angle α i , as shown in Figure 3. The transmission axis of the analyzer A is tunable by free rotating, within an angle range 0 -α. The incident linearly polarized beam, by passing of the LC cell, becomes the elliptically polarized. This elliptical polarized beam converts back to the linear by means of the QWP placed after the LC, however with different polarization angle with respect to the input polarization before LC, depending on the phase retardation. By means of analizer A, we measured the azimuth of this linearly polarized beam that allows the determination of the optical phase retardation of LC cell [29,30]. For this purpose the analyser was rotated to find the angle α corresponding minimum of intensity at the photodetector PD. Determination of the contact angle β To measure the contact angle β of isotropic liquid (glycerol) or two nematic LCs (E7 and MLC-6609), the simple method has been used [40]. It is based on the measurement of linear dimensions of the droplet placed to the aligning layer by means of the horizontal microscope, as shown in Figure 4a. AFM studies of the modified aligning layers In this section the AAE W φ for two types of aligning layers will be studied in detail. However, in some cases, the FTAL and STAL under studied will be also compared with The dependence of the nano-grooves depth A on the scanning speed υ at constant LPF J = 0.55 J/cm 2 is shown in Figure 6. For the pure nano-structured Ti layer this dependence is a non-monotonic function (curve 1, solid red squares) as can be seen from Figure 6. However, it is Important to note that monotonous growing of the depth of grooves A has been observed when the process of the treatment of the Ti surfaces was carried out at a constant scanning speed υ for different value of LPF J, increasing its [19]. In our experiments the average value of period Λ was about 910 nm, for the treated by NLL nano-structured Ti surfaces, which agree within 10 nm with values in [19,20]. In comparison with the pure structured Ti layers, the depth of grooves A of the Conversely, the depth of grooves A for the STAL is decreased within range 30-3, opened green triangles). It is therefore clear that the coating of the FTAL by polyimide PI2555 leads to the better filling of low-lying place of nano-grooves than their peaks. Due to this fact the depth of nano-grooves A has decreased but the period of grooves Λ is the same as for the FTAL. The estimation of AE by Berreman theory By knowing values of the depth A and period Λ of nano-grooves an estimation of AE W B by using the well-known Berreman's theory [58,59] can be carried out. The value of AE W B for the pure nano-structured Ti layer, the FTAL and the STAL was calculated as follows: where K is arithmetical mean of the all Frank constants (K 11 , K 22 and K 33 ) of nematic LC. The dependence of AE W B on scanning speed υ for the pure nano-structured Ti layer (curve 1, solid red squares), the FTAL (curve 2, solid blue circles) and the STAL (curve 3, solid green triangles) as an example for the nematic E7 are shown in Figure 7a. It is important to note that the ITO layers coating the pure nano-strucrured Ti layers (so-called FTAL) lead to the certain increasing the depth of nano-grooves A and little change of period Λ ( Figure 5 and Figure 6). The dependence of AE W B on depth of grooves A for the pure nano-structured Ti layers (opened red squares) and two modified layers, namely FTAL (opened blue circles) and STAL (opened green triangles) is shown in Figure 7b. Thus it can be concluded that some modification of pure nano-structured Ti layers such as coating by ITO and polyimide PI2555 layers can also change the depth of nano-grooves A and therefore the AE W B . As can be seen from Figure 7a and Figure 7b [19,20] we recently found the difference between the calculated AE W B and experimentally obtained, using the method of combined twist LC cell [36,37], AAE W φ ,. One of the main reasons for this difference there is the fact that Berreman's theory has been not fully incorporating the physical and chemical interactions between LC molecules and surface [19,20]. As can be seen from Figure 7b (line 1, opened blue circles), the usage of any additional coating, which may increase the depth of nano-grooves A or/ and decrease of the period of nano-grooves Λ, will increase AE W B . Determination of AAE W φ through the measurement of the twist angle φ Recently, it was shown that the AAE W φ of the treated by NLL surfaces could depend on, at the least, the two main parameters (e.g. LPS J and scanning speed υ), which use at the time of the creation of nano-structured layers [19,20]. In this manuscript the AAE W φ as a function of the scanning speed υ was also computed by twist angle φ measurements, by using well-known technique of combined twist LC cell [51][52][53][54]. The observed. The obvious conclusion is that the FTAL, consisting of the ITO layers, possess less surface tension γ S than tension of LC γ LC (i.e. γ S < γ LC ) and as assumed by Creagh-Kmetz's rule [38,39], the P-H transition occurs. As can be seen from Figure 9a, the increasing of AAE W φ is strongly related to the rise in nano-grooves depth A. These dependencies are qualitatively coincident with those calculated previously [19,20] by means of the Berreman's theory. However, as mentioned above, the value of the AAE W φ can also depend on the changing of physical and chemical properties of the surface. Here, the modification of aligning layers was realized through the usage of various materials, namely ITO and PI2555. Recently, it was concluded that the increasing of AAE W φ could be gained by the usage of polymer having stronger anchoring [19,20]. In this manuscript, in contrast to recently used polyimide ODAPI, having AAE W φ ~ 1×10 -4 J/m 2 [19,20] [62]. In Figure 9b the dependence of AAE W φ on twist angle φ is shown. This dependence is a linear function with a certain slope ψ (Fig. 9b) Similar dependencies φ(υ) and W φ (υ) for LC cells filled by the nematic MLC-6609 (Δε < 0) are shown in Figure 10. Here, in comparison to the nematic E7 for the pure nano-structured Ti layers [19,20], homeotropic alignment was observed after the filling LC cells by the nematic MLC-6609. The twist angles φ and the calculated AAEs W φ as functions of the scanning speed υ for the FTAL (curve 1) and the STAL (curve 2) are shown in Figure 10a and Figure 10b, respectively. It should also be mentioned that similarly to LC cells filled by the nematic E7 (Δε > 0), for combined twist LC cells, consisting of the FTAL (curve 1) and filled by the nematic MLC-6609, the P-H transition occurs within 2 hours. Therefore, by using FTAL (curve 1) the planar alignment of the MLC-6609 will be always unstable. Because of this finding, the LC cells are unable to rotate the linearly polarized light similarly to combined twist LC cells described elsewhere [36,37,62]. Consequently, it should be noted that in the case of FTAL, the measurements of the twist angles φ were carried out immediately after filling of LC cells. As can be seen from Figure 8 and Figure Polar anchoring energy W θ of aligning surfaces In this section the influence of the scanning speed during treatment of Ti surface and their modification by different layers (e.g. ITO and polyimide PI2555) on the pretilt angle θ p and PAE W θ of symmetrical LC cells would be investigated. The LC cell was consisted of a pair of substrates coated with FTAL or STAL and filled by nematics E7 (Δε > 0) or MLC-6609 (Δε < 0). It is well known that for the calculation of PAE W θ , different methods [27][28][29][30][31]33] can be used. The methods proposed involve studying of LC cells under the action of an alternating electric field. As mentioned above in the Introduction section, because of this fact, the LC cells consisting of pair pure nano-structured Ti layers will not be examined, because these LC cells not conduct of the current over the whole treated surface area. The measurement of pretilt angle θ p In Figure 11 In order to calculate the PAE W θ , two different methods of the optical phase retardation measurements were used [29,30,33]. To measure of optical phase retardation of FTAL the optical scheme for VA of LC cell filled by nematic MLC-6609 with negative Δε < 0 was used [33]. In case of STAL the simple Senarmont compensator [56] was used in order to measure of the optical phase retardation of the LC cell possessing the planar alignment as detailed previously [29,30,32]. The typical optical phase retardation function R(V-V * ) of LC cells from applied voltage (V-V * ), which was used to calculate PAE W θ of aligning layers is shown in Figure 12. As can be seen from Figure 12, in the certain voltage range (V min , V max ) the optical retardation as a function of applied voltage is a linearly dependent ). (Figure 13a, curve 2) due to the polyimide PI2555 layer, characterized by strong AAE W φ [60,61]. Let us recall, a strong AAE there is the reason for decreased pretilt angle θ p (Figure 11a, curve 3). As can be seen from Figure 13b the increasing of pretilt angle θ p leads to PAE W θ growth. The inclination angle  of linear dependence W θ on θ p may be seen as a typical parameter, which characterizes a certain type of aligning layer. In addition, under the certain experimental conditions some correlation between PAE W θ and AAE W φ is observed. In contrary to [63], here for example, nanostructured surfaces possessing relatively strong PAE W θ (Figure 13a, curve 1) have a weak AAE W φ (Figure 8b, curve 2), and vice versa. The same correlation between AAE W φ and PAE W θ were recently found during their measurements and studies of easy orientation axis gliding in a magnetic field on photoaligning PVCN-F layer [64]. Here authors found that relatively weak AAE W φ = (10 -7 -10 -5 ) J/m 2 leads to strong drift of easy orientational axis when PAE W θ = 10 -4 J/m 2 in a strong electric field. As can be seen from Figure 13a, the coating of the PI2555 layer onto the FTAL leads to weaker PAE W θ of the formed STAL (curve 2). Here, as already mentioned above, the main reason of stronger AAE W φ of STAL (Figure 8b, curve 3) is polyimide PI2555 layer, possessing strong AAE W φ in comparison with FTAL ( Figure 8b, curve 2). This correlation between PAE W θ and AAE W φ can also be understood from the schematic presentation of distribution of the director n  in the bulk of LC, which is bordered on the nano-structured aligning surface (Figure 1). The dependence of PAE W θ on AAE W φ both FTAL (opened blue diamonds) and STAL (solid green spheres) is also shown in Figure 14, when LC cells were filled by the nematic MLC-6609 (Δε < 0) and E7 (Δε > 0), respectively. It is seen that under certain conditions, namely when PAE W θ of the aligning layer is increasing (or pretilt angle θ p of the nematic MLC-6609 is decreasing) the decreasing of AAE W φ of this layer is observed. For instance, in the case of various FTAL (opened blue diamonds), which differ by scanning speed υ during the NLL processing, VA of the nematic MLC-6609 with small pretilt angle θ p (close to 90 degrees) occurs. As can be also seen from Figure 14, for the STAL (solid green spheres), having the polyimide PI2555 layers, the strong AAE W φ is observed. This results in planar alignments of the nematic E7. Contact angle β of nematic droplets on the aligning layers It is assumed that one main reason for wettability of solid surfaces is both the interaction between interfacial tension of all materials which make contact and the geometrical properties of the surface (i. e. random roughness or periodical nano-grooves with period Λ and depth A). By taking into account of the empirical Creagh-Kmetz's rule [38,39], here, some correlation between the contact angle β of nematic droplets deposited on FTAL and STAL, and calculated AAE W φ (or/and PAE W θ ) will be described. Contact angle of the droplets on rubbed polymer films First of all, in order to demonstrate the impact of both various polymers and different processing methods on contact angle β of droplets (e.g. two nematics and glycerol), we have chosen the polyimide PI2555 and ODAPI, having different values of AAE W φ (PAE W θ ) [60][61][62]. At the beginning, the dependence of the droplets' contact angle β for two nematics and glycerol on the number of rubbing N rubb for the polyimide PI2555 and ODAPI layers was studied. As can be seen from Figure 15, for droplets of the two nematics and glycerol the non-monotonic dependence β(N rubb ) for each polymer layers is observed. In this case all the measurements of geometrical dimension of droplets were carried out along direction of rubbing. In the case of ODAPI, a certain correlation between the contact angle β (Figure 15a, curve 2) and calculated value of AAE W φ [62] for the number of rubbing N rubb is observed, similarly to the case of the photosensitive PVCN-F surface [40]. Consequently, for the ODAPI layers possessing a strong AAE W φ (when N rubb = 10 -15), small contact angles β were found. As can be seen from Figure 15a for the glycerol (curve 1, opened red squares) and two nematics, namely the E7 (curve 2, opened blue circles) and the MLC-6609 (curve 3, opened green triangles), the contact angle β increasing for ODAPI layers having a weak W φ . Contact angle of droplets on modified nano-structured surfaces It may be assumed that the aligning layer created by NLL and further coated by ITO and The dependence of the contact angle β for glycerol and two nematics droplets, deposited on both FTAL and STAL, on scanning speed υ is shown in Figure 17a and As can be seen from Figure 17, for all liquids, deposited on the FTAL, contact angles β are more than in case of STAL under the same conditions of treatment of Ti surface by means of NLL. It can be assumed that FTAL has low wettability of the surface than STAL, because STAL consists of the polyimide PI2555 layer possessing strong AAE W φ . As noted above, the aligning layer, having strong AAE W φ , characterized by weak PAE W θ and vice versa [64]. It is obvious that certain correlation between the contact angle β and AAE W φ (and/ or PAE W θ ) may also be seen. Dependence of the calculated AAE W φ (lines 1) and PAE W θ (lines 2) on the contact angle β for nematic droplet MLC-6609 deposited on FTAL and nematic droplet E7 deposited on STAL are shown in Figure 18a and Figure 18b, respectively. It is easy to see that the decreasing of AAE W φ (lines 1) resulting in the increasing of the contact angle β for each nematic droplets when using both FTAL ( Figure 18a) and STAL (Figure 18b). In addition, for FTAL, having a stronger PAE W θ (Figure 18a, line 2 and opened blue diamonds), the contact angle β is more than for STAL, possessing a weak PAE W θ (Figure 18b, line 2 and solid blue diamonds). Conclusions In this manuscript, for the first time, we studied in detail the aligning surfaces, obtained by the NLL method, which was recently proposed [19,20] ). It was found that the certain correlation between the contact angle β of nematic droplet deposited on the FTAL/STAL and scanning speed υ during the process of NLL and therefore between the contact angle β and PAE W φ and AAE W θ take place.	2020-03-24T01:01:09.661Z	2020-03-21T00:00:00.000	{ "year": 2021, "sha1": "316eefee697d6595eae662c5f6b449b392afbc8e", "oa_license": null, "oa_url": "http://arxiv.org/pdf/2003.09706", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "316eefee697d6595eae662c5f6b449b392afbc8e", "s2fieldsofstudy": [ "Physics", "Materials Science", "Engineering" ], "extfieldsofstudy": [ "Physics", "Materials Science" ] }
4476009	pes2o/s2orc	v3-fos-license	Joint Causal Inference from Observational and Experimental Datasets We introduce Joint Causal Inference (JCI), a powerful formulation of causal discovery from multiple datasets that allows to jointly learn both the causal structure and targets of interventions from statistical independences in pooled data. Compared with existing constraint-based approaches for causal discovery from multiple data sets, JCI offers several advantages: it allows for several different types of interventions in a unified fashion, it can learn intervention targets, it systematically pools data across different datasets which improves the statistical power of independence tests, and most importantly, it improves on the accuracy and identifiability of the predicted causal relations. A technical complication that arises in JCI is the occurrence of faithfulness violations due to deterministic relations. We propose a simple but effective strategy for dealing with this type of faithfulness violations. We implement it in ACID, a determinism-tolerant extension of Ancestral Causal Inference (ACI) (Magliacane et al., 2016), a recently proposed logic-based causal discovery method that improves reliability of the output by exploiting redundant information in the data. We illustrate the benefits of JCI with ACID with an evaluation on a simulated dataset. Introduction Discovering causal relations from data is at the foundation of the scientific method. Traditionally, causal relations are either recovered from experimental data in which the variable of interest is perturbed, or from purely observational data, e.g., using the seminal PC and FCI algorithms (Spirtes et al., 2000;Zhang, 2008). In recent years, several methods for combining observational and experimental data to discover causal relations have been proposed, showing that this combination can improve greatly on the accuracy and identifiability of the predicted causal relations. Some of the proposed methods are score-based (e.g., Cooper and Yoo, 1999;Tian and Pearl, 2001;Eaton and Murphy, 2007;Hauser and Bühlmann, 2012;Mooij and Heskes, 2013), i.e., they evaluate models using a penalized likelihood score, while others (e.g., Tian and Pearl, 2001;Claassen and Heskes, 2010;Hyttinen et al., 2014;Triantafillou and Tsamardinos, 2015;Peters et al., 2015;Magliacane et al., 2016;Borboudakis and Tsamardinos, 2016) are constraint-based, i.e., they use statistical independences to express constraints over possible models. In this work we propose Joint Causal Inference (JCI), a formulation of causal discovery over multiple datasets in which both the causal structure and targets of interventions are jointly learnt from independence test results in pooled data. A related approach was already proposed for scorebased methods by Eaton and Murphy (2007), but here we extend it so that constraint-based methods can be employed. Our goal is to combine the idea of joint inference from observational and experimental data from Eaton and Murphy (2007) with the advantages that constraint-based methods have over score-based methods, namely, the ability to handle latent confounders and selection bias naturally in a nonparametric approach, and, especially in the case of logic-based methods, an easy integration of background knowledge. Existing constraint-based methods for multiple datasets typically learn the causal structure on each dataset separately and then merge the learnt structures (e.g., Claassen and Heskes, 2010;Hyttinen et al., 2014;Triantafillou and Tsamardinos, 2015;Borboudakis and Tsamardinos, 2016). The merging process depends on the type of interventions, and most existing methods support only interventions on known targets. Instead, JCI: (1) allows for several different types of interventions; (2) can learn the intervention targets; (3) systematically pools data across different datasets, which improves the statistical power of independence tests; and (4) improves the identifiability and accuracy of the predicted causal relations. On the other hand, JCI poses challenges for current constraint-based methods because of their susceptibility to violations of the Causal Faithfulness assumption. Specifically, JCI induces faithfulness violations due to deterministic relations, which would typically result in erroneous inferences with standard constraint-based methods. We propose a simple but effective strategy for dealing with this type of faithfulness violations. The strategy can be applied to any constraint-based causal discovery method for observational data that can handle partial inputs, i.e. missing results for a certain independence test, thus extending it to a JCI method that can handle a combination of observational and experimental data. We implement the strategy in ACID (Ancestral Causal Inference with Determinism), a determinism-tolerant extension of ACI (Ancestral Causal Inference) (Magliacane et al., 2016), a recently proposed logic-based causal discovery method that improves reliability of the output by exploiting redundant information in its input. In our evaluation on synthetic data we show that JCI with ACID improves on the accuracy of the causal predictions with respect to simply merging separately learned causal graphs, illustrating the advantage of joint causal discovery. Preliminaries In this section we review a few useful concepts from the related work and introduce the notation we use in the rest of the paper. Most of the concepts described here are explained in detail in the seminal books by Pearl (2009) and Spirtes et al. (2000). In the following, we represent variables with uppercase letters, while sets of variables are denoted by boldface. Graph terminology Throughout the paper we assume that the data generating process can be modeled by a causal Directed Acyclic Graph (DAG) that may contain latent variables. For simplicity, we do not consider selection bias. A directed edge X → Y in the causal DAG represents a direct causal relationship of cause X on effect Y . We then say that X is a parent of Y , and denote the set of parents of Y as PA(Y ). A sequence of directed edges X 1 → X 2 → · · · → X n is a directed path. If there is a directed path from X to Y (or X = Y ) then X is an ancestor of Y (denoted as X Y ). We denote the set of ancestors of Y as AN(Y ). If there is no directed path from X to Y (and X = Y ) we denote this as X Y . A sequence of unique nodes X 1 , . . . , X n such that each pair {X i , X i+1 } is connected by an edge is called a path. A collider is a node on a path that has two incoming arrow heads from both its neighboring nodes on the path, i.e., of the form X i−1 → X i ← X i+1 . Any other node on a path (including the end nodes X 1 , X n ) is called a non-collider. For a set of variables W , we extend the definition of parents PA(W ) to the union of all parents of any variable W ∈ W . We similarly define AN(W ) as the set of ancestors of all W ∈ W . We write X W if there exists at least one effect Y ∈ W that has X as an ancestor, i.e., X Y . Independences and d-separation For disjoint sets of random variables X, Y , W , distributed according to a joint probability distribution P, we denote the conditional independence of X and Y given W in P as X ⊥ ⊥ Y \| W [P], and conditional dependence as X ⊥ ⊥ Y \| W [P]. We often omit [P] when it is obvious from the context which probability distribution we are referring to. We call the cardinality \|W \| the order of the conditional (in)dependence relation. A well-known graphical criterion for DAGs, with many implications for causal discovery, is d-separation (Pearl, 2009;Spirtes et al., 2000): , iff every path π in G that connects any X ∈ X with any Y ∈ Y is blocked by W , i.e. at least one of the following holds: • π contains a collider not in AN(W ), or • π contains a non-collider in W . The opposite, i.e., d-connection, is denoted as . We often omit [G] from the notation when it is obvious which DAG we are referring to. The following two key assumptions for constraint-based causal discovery have been thoroughly discussed in the literature (see for example Spirtes et al., 2000). They connect conditional independences in the observational distribution P with d-separations in the underlying causal DAG G. • Causal Markov Assumption: d-separation in the causal DAG G implies conditional independence in the observational distribution P. For all disjoint sets of variables X, Y , W : which can also be expressed contrapositively as: In the left example, we show a set of faithfulness violations due to a functionally determined relation, in which the parent X fully determines the child Y . In the right example, X and S are binary variables and ⊕ is the XOR function. Conditioning on {S, Y } fully determines X, even though {S, Y } are not ancestors of X. This creates a series of non-trivial faithfulness violations, listed below the graph. In both graphs, we represent the variables resulting from structural equations without noise terms with a double circled node. • Causal Faithfulness Assumption: the inverse, i.e., for all disjoint sets of variables X, Y , W : If we assume both the Causal Markov and Causal Faithfulness assumptions to hold, the conditional independences of the observational distribution correspond one-to-one with the d-separations in the causal DAG. This setting is very favourable for causal discovery, thus both assumptions are usually made in constraint-based approaches. Deterministic relations and faithfulness violations Although often reasonable, the Causal Faithfulness assumption is violated in some cases, notably in the common case of deterministic relations among variables, e.g., for Structural Causal Model equations (Pearl, 2009) in which there is no noise term. Some of the faithfulness violations related to determinism are captured by an extension to the d-separation criterion, the D-separation criterion, first introduced by Geiger et al. (1990) and later extended by Spirtes et al. (2000). Under the Causal Markov assumption, the formulation of D-separation presented by Geiger et al. (1990) is proven to be complete for the restricted setting where determinism arises only due to functionally determined relations, defined recursively as variables that are fully determined by their parents, or more precisely: Definition 2 A variable X is functionally determined by a set W for a given DAG G if X ∈ W , or all parents of X are functionally determined by W . We show an example of faithfulness violations due to a functionally determined relation in Figure 1 (left). Following Geiger et al. (1990), we represent the variables resulting from structural equations without noise terms with a double circled node. In Section 3.8 of their book, Spirtes et al. (2000) extend D-separation to model also some deterministic relations that are due to variables being determined when conditioning on their nonancestors. In Figure 1 (right) we show an example in which the definition of D-separation from Geiger et al. (1990) fails to capture the faithfulness violation that is due to the deterministic relation between non-ancestors of X, but the extended notion of D-separation by Spirtes et al. (2000) correctly captures it. Although the version of D-separation by Spirtes et al. (2000) retains completeness for the restricted case of functionally determined relations, it is not proven to be complete in general. Nevertheless, Spirtes et al. (2000) introduce several useful concepts for handling general deterministic relations, so we summarize their findings here, adapting them to our notation. We start with the assumption that we have complete knowledge of all deterministic relations in the system. This assumption is not as restrictive as it may seem at first, because in practice one can easily reconstruct deterministic relations in the data by using several standard methods. We use D to define a function that maps a given set of variables W to the set of all variables that are determined by W : Definition 3 Given a set of variables W and a complete set of deterministic relations D, we define DET D (W ) as the set of variables determined according to D by (a subset of) W . We omit D, using only DET(W ), if it is obvious from the context which set we are referring to. Note that DET D (W ) trivially includes W itself. Also, any variable with constant value is by definition in DET D (W ) for all W as it is determined by ∅. Following Spirtes et al. (2000), we can use DET() to extend d-separation for deterministic relations. Definition 4 Given a DAG G, three disjoint sets of variables X, Y , W , and the complete set of deterministic relations D, we define X and Y to be D-separated by W w.r.t. D and G, denoted as X ⊥ D Y \| W [D, G], iff for every path π in G between any X ∈ X and any Y ∈ Y , at least one of the following holds: • π contains a collider not in AN(W ), or • π contains a non-collider 1 in DET D (W ). If DET D (W ) = W , D-separation reduces to standard d-separation. We omit D and G if it is obvious from the context which set and graph we are referring to. Under the Causal Markov Assumption, this formulation of D-separation is proven to imply independence (Spirtes et al., 2000). More precisely: if P is Markov with respect to G and D is the complete set of deterministic relations that hold in P. For the case of functionally determined relations, this version of D-separation is complete, i.e., it completely identifies all additional independences due to functionally determined relations, because in that setting it reduces to the version of D-separation by Geiger et al. (1990), which was shown to be complete by Geiger et al. (1990). Related work Given a set of observational and interventional datasets, most constraint-based methods that handle multiple datasets learn the causal structure from each dataset separately and then merge the learned structures (e.g., Claassen and Heskes, 2010;Hyttinen et al., 2014;Triantafillou and Tsamardinos, 2015;Borboudakis and Tsamardinos, 2016). Some of these methods, e.g., COMBINE (Triantafillou and Tsamardinos, 2015), perform these two steps sequentially, applying a greedy procedure to resolve any potential conflict from the causal structures learnt in the first step. Others, e.g., HEJ (Hyttinen et al., 2014), combine learning and merging in a single procedure, solving potential conflicts by formulating an optimization problem. A recent approach, ETIO (Borboudakis and Tsamardinos, 2016), combines the aspects of the previous approaches by learning and merging in a single procedure, but using a greedy algorithm for resolving conflicts. Merging causal structures learnt on each dataset separately has several drawbacks with respect to a method that can jointly use all datasets, as for example the score-based method from Eaton and Murphy (2007). First, merging approaches require known targets and cannot learn the targets of the interventions, since this type of information is only available when considering multiple datasets jointly. Moreover, they cannot take advantage of certain interventional datasets, e.g., in the case of a single data point per interventional setting, as for example happens in a popular genomics dataset (Kemmeren et al., 2014). Two other important drawbacks are a loss of statistical power because of the separation into smaller isolated datasets, and, as we will show with some examples in Section 4.1, less identifiable relations with respect to a joint causal inference method. There is some related work on special cases in which to apply constraint-based methods with mixtures of observational and experimental datasets (e.g., Tian and Pearl, 2001;Eberhardt, 2008;Lagani et al., 2012;Peters et al., 2015;Borboudakis and Tsamardinos, 2016), but the problem has not been systematically discussed and formalized in a general framework yet. In particular, to the best of our knowledge, no existing work addresses learning the intervention targets, possibly jointly with the causal graph, from independence tests. The approach described by Tian and Pearl (2001) learns the causal structure by combining a standard constraint-based method on observational data with information extracted from changes in the marginal probability of each variable. This information can be extracted from a sequence of (interventional) datasets by comparing each pair of datasets in the sequence, under the assumption that the only difference between the two datasets is a mechanism change on a single known variable. This is a quite restrictive assumption in practice. Other approaches describe sufficient, although restrictive, conditions under which pooling data does not change the conditional distribution of the variables under consideration. In particular, Eberhardt (2008) describes how naively pooling data from different experimental settings, while discarding the information of which experimental setting a sample was taken, may give wrong results. Thus Eberhardt (2008) proposes a sufficient condition that allows one to pool data for a given independence test when the conditional distribution of the tested variables is the same in all experimental conditions. Lagani et al. (2012) present two other approaches: (i) perform the conditional independence tests separately in each dataset, then define the pooled dependence as a disjunction of the single dependences; (ii) pool experimental conditions that differ only in the value of at most one intervened variable. Both of these approaches describe restrictive conditions in which one can pool datasets, while in this paper we argue that, when done systematically, e.g., as we will show in the next section, one can always pool all available datasets. Other approaches like Invariant Causal Prediction (ICP) (Peters et al., 2015) focus on certain specific combinations of independence tests that are performed jointly on all datasets. ICP is a causal discovery method that looks for invariance across different experimental settings, returning a conservative subset of ancestors (or parents, if one assumes there are no latent confounders) for a given target variable Y . The main assumption is that the conditional distribution of Y given its parents does not change in the different interventional settings (in particular, that Y is not directly intervened upon). This assumption is also referred to as invariance or modularity (Pearl, 2009;Spirtes et al., 2000). Since the method searches for patterns that are invariant across different settings, it can safely pool together a subset of settings in a new virtual "experimental" setting to increase the statistical power for settings with few data. On the other hand, as we show with some examples in Section 4.2, the conservativeness of the ICP estimates sometimes significantly reduces the causal information that can be inferred. Like ICP, JCI makes an invariance assumption that allows it to combine different datasets. The invariance assumption made by JCI is that the causal structure is invariant across experimental settings (but model parameters are allowed to change). Joint Causal Inference (JCI) We propose to model jointly with a single causal graph n observational or experimental datasets {D r } r∈{1,...,n} . We assume that there is a unique underlying causal DAG G X in all of these datasets, defined over the same set of variables that we call the system variables, {X j } j∈X , some of which are possibly hidden. Each dataset D r has an associated joint probability distribution P r ((X j ) j∈X ) and represents the d r data points collected after a set of interventions on possibly unknown intervention targets. In the context of this paper, observational data are simply datasets with an empty set of interventions. We assume each distribution P r (r = 1, . . . , n) to be Markov and faithful with respect to the causal DAG G X . This assumption precludes certain types of interventions, notably, perfect interventions (Pearl, 2009). On the other hand, it allows for many other types of interventions, e.g., soft interventions (Markowetz et al., 2005), mechanism changes (Tian and Pearl, 2001), fat-hand interventions (Eaton and Murphy, 2007), activity interventions (Mooij and Heskes, 2013), etc., as long as they do not induce new (in)dependences, which can be seen as modifications to the underlying DAG G X . Using the terminology from Dawid (2002), we call the different distributions in the datasets regimes. In related work different names have been used, e.g., experimental conditions or environments (Mooij and Heskes, 2013; Peters et al., 2015). We introduce two types of dummy variables in the data: • a regime variable R, representing which dataset D r a data point is from, i.e., ∀r = 1, . . . , n, R = r for data from D r . • intervention variables {I i } i∈I , which are deterministic functions of the regime R. Intervention variables represent the interventions performed in each dataset. In absence of any information on the interventions performed in the datasets, we can use as intervention variables the indicator variables for each of the datasets. We can now state the main assumption of JCI. Assumption 2 We assume that the causal relations between system variables {X j } j∈X and the introduced dummy variables R and {I i } i∈I can be represented as an acyclic Structural Causal Figure 2: Prototypical example of JCI setting: A set of four experimental datasets in raw form (left), in a pooled tabular form with the addition of dummy variables (center) and as a causal DAG G (right) representing the causal structure of the system variables X 1 , . . . , X 4 , regime variable R and intervention variables I 1 , I 2 . The intervention variable I 1 represents the temperature at which each experiment was performed, while I 2 represents the addition of a drug in some of the experiments. Model (SCM) M with jointly independent exogenous variables Here, PA X (X j ) are the system variable parents of X j , while PA I (X j ) denote its intervention parents and E j is the exogenous parent of X j . The distribution P r corresponding to dataset D r is given by p((X j ) j∈X \| R = r). We denote the corresponding causal DAG as G. The Causal Markov assumption then holds by construction. We show an example in Figure 2, where we model four datasets with the same underlying causal structure. The JCI assumptions are applicable when the value of the regime/intervention variables are determined by the experimenter before the system variables are measured. More generally, the assumption is that the system variables cannot cause the regime/intervention variables. In addition, we assume that the values set by the experimenter are chosen independently of any other possible cause of the system variables. In other words, we assume there to be no confounders between the regime variable and the system variables, or between the intervention variables and the system variables. For the purposes of causal discovery as intended in this paper, there is nothing else that really distinguishes the regime/intervention variables from the system variables: they can both be considered to be random variables, where the distribution of the regime/intervention variables just reflects the empirical distribution of the experimental design chosen by the experimenter. Moreover, there is no distinction between observational and experimental datasets, allowing for several observational datasets, possibly from different contexts. Table 1: Example of experimental design matrix with additional deterministic relations beyond the ones allowed in JCI (left) and a reduced version with only allowed deterministic relations (right). In the right version we joined I Akt-Inh and I U0126 in a single intervention variable I drug representing the addition of a single drug (U0126 when the value is 0, Akt-Inh when it is 1). Intervention variables are functions of the regime variable, and do not have any associated noise. This means that they are determined by the regime. We represent these functions as a matrix: Definition 5 We define the experimental design matrix as the matrix representing the functional relations between R and each intervention variable I i , and the corresponding probabilities of the regime variable p(R = r) = dr Σ s∈{1,...,n} ds , where d s is the number of data points in dataset D s . We assume that the intervention variables are complete in the sense that every effect of the regime variable is mediated through an intervention variable. In other words, we assume that there are no direct effects of R on any of the system variables. In general, other deterministic relations between dummy variables may arise. For example, consider the example in Table 1 left in which I Akt-Inh represents a drug that was added when the regime is an odd number, while I U0126 indicates another drug that was added when the regime is an even number. These two variables determine each other. Even though this is clear from the experimental design matrix, it is not visible in the causal influence diagram. In this paper, we focus on a special case by allowing only certain types of deterministic relations. We assume that the regime R determines each of the intervention variables {I i } i∈I . Optionally, we allow one additional deterministic relation, namely that all intervention variables {I i } i∈I together determine the regime R. We assume that there are no other deterministic relations. Assumption 3 The deterministic relations that hold in the joint distribution P(R, {I i } i∈I , {X j } j∈X ) are {R}, I i for all i ∈ I, and optionally, {I i } i∈I , R . No other deterministic relations hold in the joint distribution. In practice, one can often "normalize" a system that does not satisfy this assumption. For example, the experimental design matrix in Table 1 left contains also deterministic relations that are not allowed in JCI, but that arise from "redundant" intervention variables. Table 1 right shows how joining two intervention variables can yield a "normalized" system that satisfies the JCI assumptions. D-separation has been shown to be sound (Spirtes et al., 2000), but was only conjectured to be complete. We prove that in our restricted setting, D-separation is actually complete. Theorem 6 Under Assumptions 1-3, D-separation is complete, i.e., it gives all conditional independences that are entailed by the assumptions. Proof First we consider the case that the regime variable R is not determined by the intervention variables. Then, all deterministic relations are functionally determined relations (see Definition 2), as they correspond to each intervention variable being a function of the regime variable only. The notion of D-separation introduced by Geiger et al. (1990) is proved to be sounds and complete under the Causal Markov assumption for functionally determined relations. If all deterministic relations are functionally determined relations, then their notion of D-separation is equivalent to the one by Spirtes et al. (2000) that we use here. Thus the statement follows. For the other case, we will show that the D-separations do not change by removing the deterministic relation that the regime variable R is determined by all intervention variables {I i } i∈I . Let D denote the complete set of deterministic relations according to M and assume {I i } i∈I , R ∈ D. Let D * = D \ {I i } i∈I , R . Let G be the DAG associated with M. We claim that for disjoint sets of variables X, Y , W : G]. This can only happen when {I i } i∈I ⊆ W , because otherwise DET D (W ) = DET D * (W ) and then the two D-separations are identical by definition. So there must exist a path π in G that is D-open w.r.t. D * but D-closed w.r.t. D. This means it must contain R as a non-collider. Also, R / ∈ W otherwise the path would be closed w.r.t. D * . Since X and Y are disjoint, π must contain at least one other node that is adjacent to R, which must be one of the intervention variables {I i } i∈I . Since the intervention variables can only be non-colliders on π by the JCI assumptions, and {I i } i∈I ⊆ W , they must d-block π. Hence we have arrived at a contradiction: π cannot be D-open w.r.t. D * . Moreover cannot happen, because d-separations are the same for D and D * , and D * ⊂ D, so there cannot be additional D-separations w.r.t. D * compared to the D-separations w.r.t. D. Therefore, removing this particular deterministic relation does not change the D-separation statements, and hence completeness follows also for this case. We conjecture that for the more general case of arbitrary deterministic relations between dummy variables, e.g., Table 1 left, D-separation as defined by Spirtes et al. (2000) is still complete, but we leave the proof for future work. The completeness of D-separation is important in the JCI context because it motivates a relaxation of the standard Causal Faithfulness assumption. In our setting, the standard assumption is too restrictive, so we relax it to allow for violations due to deterministic relations between the regime and the intervention variables. We define our relaxed version, that we call D-Faithfulness assumption, as follows: Assumption 4 (D-Faithfulness) For three disjoint sets of variables X, Y , W and a probability distribution P that satisfies both the Causal Markov assumption for G and the set of deterministic relations D, we assume that This assumption, in conjunction with the previous ones, implies that in JCI independences correspond one-to-one with D-separations, which paves the road for constraint-based causal discovery 2 . The completeness of D-separation suggests that this relaxation is "tight", i.e., we only relax the standard Causal Faithfulness assumption to allow for the extra independences that are due to the deterministic relations in Assumption 3, but not any more. 2. A consequence of D-Faithfulness is that intervention variables should be pairwise dependent (also when conditioning on a subset of them). This excludes some experimental design matrices, e.g. a matrix similar to Table 1, but where P (R = r) = 0.25 for all r. In practice, we can often alleviate this problem, e.g. by dropping some data points. Figure 3: A simple example in which JCI improves identifiability: Consider two datasets with the same underlying DAG, one of which has a soft intervention (left). If we learn the causal graphs of X 1 and X 2 in each dataset separately and then merge them, e.g., as described by Triantafillou and Tsamardinos (2015), we cannot learn the causal direction but only that they are dependent (middle). JCI adds more variables and thus conditional independence tests that allow to distinguish the direction (right). Here, since R = I 1 , w.l.o.g. we represent only I 1 . See details in the main text. Under Assumptions 1-4, we define Joint Causal Inference (JCI) as the problem of inferring the causal DAG G from the distributions (P r ) r=1,...,n or from finite samples (D r ) r=1,...,n of those. Moreover, we call any causal discovery method that can solve a JCI instance a JCI method. We will show in Section 4.2 that the ideas behind some previous approaches, e.g., (Peters et al., 2015), can be seen as special cases of JCI. Joint Causal Inference improves on the identifiability w.r.t. merging learnt structures As already mentioned, formulating causal inference on multiple datasets as JCI offers several advantages with respect to the approaches in which the causal graphs are learnt separately from each dataset and then merged. One of the advantages is the improved identifiability. In this section, we show a few examples, where, for simplicity, we assume oracle inputs and model only the regime variable (rather than the regime variable and a single intervention variable). In Figure 3 we show a simple example in which JCI improves identifiability. In the absence of information on the intervention targets, we cannot identify the causal direction between the variables when we learn the structures separately and then combine them. In the same case, a JCI method is able to correctly reconstruct the causal structure by using additional conditional independence tests with the intervention variable, specifically, I 1 ⊥ ⊥ X 1 , I 1 ⊥ ⊥ X 2 , I 1 ⊥ ⊥ X 2 \| X 1 , I 1 ⊥ ⊥ X 1 \| X 2 and X 1 ⊥ ⊥ X 2 \| I 1 . Using the background knowledge from JCI that system variables cannot cause the intervention variable I 1 , i.e., X 1 I 1 ∧ X 2 I 1 , and that there are no latent confounders between I 1 and the system variables, we can infer I 1 → X 1 , X 1 → X 2 and that there are no confounders with any JCI method supporting direct causal relations. Note that in this example, since there are only two datasets, R = I 1 , so for simplicity we represent only I 1 . If for each datasets the targets of the intervention are known, then it is possible to retrieve their descendants (and non-descendants) by checking which variables change in each interventional dataset with respect to the observational case. This technique was successfully applied in (Magliacane et al., 2016) to retrieve a list of weighted ancestral relations that could be used as background Ancestral relations: Merged PAG (using ancestral relations): JCI method Figure 4: A more complex example in which JCI improves identifiability: If we have background knowledge on the intervention targets, e.g., we know that in one of the datasets X 1 and X 3 are intervened upon (left), we can use this information to extract some extra background knowledge in the form of ancestral relations. Merging separately learnt causal structures and this extra knowledge (as done e.g. by Magliacane et al., 2016) is still not enough information to recover the causal structure (middle). Instead, a JCI method can identify the true causal structure, more precisely the ADMG (right). Since R = I 1 , we can represent only I 1 . See details in the main text. knowledge. For example, if one were to know that X 1 is the intervention target in D 2 in the example in Figure 3, the change in X 2 in D 2 with respect to D 1 would imply that X 1 X 2 . Although (weighted) ancestral relations help in simple cases as the previous example, in general they cannot reproduce the same results as JCI. An example is given in Figure 4. In this case, knowing that in dataset D 2 the intervention targets are X 1 and X 3 , and observing that X 2 changes significantly, allows us to reconstruct that one of these targets causes X 2 , which is not enough to reconstruct the causal graph by merging the PAGs learnt from each dataset separately. Instead, a JCI method that supports direct causal relations can take advantage of the additional conditional independence tests with the regime variable and infer the complete DAG: Proposition 7 In the example in Figure 4, a JCI method that supports direct causal relations can reconstruct correctly the underlying causal graph, more precisely the acyclic directed mixed graph (ADMG), from oracle independence test results. Proof For readability, we provide the proof in the Appendix. Reformulation of related work as special cases of JCI Local Causal Discovery (LCD) (Cooper, 1997) is a simple algorithm that searches for variables X, Y, W that satisfy the pattern W → X → Y , where W is a variable not caused by any other variable under consideration. We can apply LCD to multiple observational and experimental datasets with soft interventions by using R as W , since the regime variable is by assumption not caused by any other variable. Then LCD can be summarized as: Figure 5: Two examples in which ICP is overly conservative in the JCI setting: In the left example, for variable Y ICP finds several sets S that satisfy R ⊥ ⊥ Y \| S, e.g., {X 1 }, {X 2 } and {X 1 , X 2 }, but their intersection is ∅. Instead, a JCI method that supports direct causal relations will correctly infer that the single parent of Y is X 2 . In the right example, naively adding intervention variables does not allow to estimate the ancestors of a variable Y , which would otherwise be estimated correctly. This rule can be seen as a restricted case of the JCI setting, in which we can iteratively pick pairs of variables (X, Y ) and apply the above rule to detect a subset of the causal graph. An extension of this approach is used in Invariant Causal Prediction (ICP) (Peters et al., 2015). ICP also considers the regime variable R (which is called the discrete environment variable in that work), but it does not model the intervention variables (see Peters et al., 2015, Appendix). Given a target variable Y that is not directly intervened upon, we can reformulate the main idea behind ICP as the search for the intersection of all the sets S such that R ⊥ ⊥ Y \| S: (1) In the absence of confounders, S * is a conservative estimate of a subset of the parents of Y , even when the Causal Faithfulness assumption is violated. If we cannot exclude the presence of confounders, as in the JCI setting, then ICP requires the Causal Faithfulness assumption to provide S * as an estimate of a subset of the ancestors of Y . We can see this reformulation of ICP as a special case of JCI that extends LCD with a more conservative estimate. In principle, one could easily integrate the conservative estimate (1) in a JCI method to provide more accurate estimates for the top predictions, but we leave this for future work. On the other hand, depending on the set of interventions in the available datasets, ICP may be overly conservative compared to JCI. Besides the restriction on the variable Y not to be directly intervened upon in any dataset, which is not necessary in JCI, there are some other cases in which ICP provides an overly conservative estimate of the set of ancestors. We show two examples in Figure 5. Specifically, in the left example the estimated set of ancestors for a variable Y that is two hops away from the intervened variable X 1 is empty, while a JCI method that supports direct causal relations can find the correct parent set {X 2 }. Similarly, as shown in the right example, naively adding the intervention variables reduces the applicability of ICP to only estimate ancestors of variables that are directly intervened upon (e.g., X 2 ). This naive addition would allow ICP to learn the intervention targets, but not the structure of the causal graph. A strategy for extending constraint-based methods for JCI Joint Causal Inference provides some challenges for current constraint-based methods: • faithfulness violations due to deterministic relations between the dummy variables, • the availability of complex background knowledge (i.e., not limited to the presence/absence of edges or ancestral relations) on the dummy variables that can improve structure learning and recover from some of the faithfulness violations (e.g., R can only cause a system variable through an intervention variable). There is some work on dealing with faithfulness violations in the PC algorithm (Lemeire et al., 2012), but it assumes causal sufficiency (in our context, no hidden variables in G), and cannot handle background knowledge. Logic-based constraint-based algorithms, (e.g., Hyttinen et al., 2014;Triantafillou and Tsamardinos, 2015;Magliacane et al., 2016) can handle complex background knowledge and causal insufficiency, but the existing implementations cannot deal with faithfulness violations due to deterministic relations. Here we propose a simple but effective strategy for dealing with faithfulness violations due to deterministic relations. We rephrase the constraints of a constraint-based algorithm in terms of dseparations and d-connections, instead of independence test results. At testing time we decide for each independence test result which d-separation or d-connection can be soundly derived from it and provide these d-separations and d-connections as input to the modified constraint-based algorithm. Before introducing the rules that we use to derive sound d-separations and d-connections from input independence test results, we first summarise the basic properties of conditional independence originally introduced by Dawid (1979), which we will use to prove an intermediate lemma. We follow the notation and ordering from a more recent publication (Constantinou and Dawid, 2015): Proposition 8 Let X, Y, Z, W be random variables. We write W Y to denote that W is a function of Y , or in other words W = f (Y ) for a measurable function f . Then the following properties hold: And for the other direction: We can now use Lemma 9 to prove a sound conversion from D-separation statements to d-separation statements: Theorem 10 For some set of variables W , let DET(W ) denote the variables determined by (a subset of) W (see Definition 3). Let X, Y be two different variables that are disjoint from DET(W ). Under the Causal Markov and D-Faithfulness assumptions, the following holds: Proof The following equivalences hold: The first equivalence follows from the Causal Markov and D-Faithfulness assumptions, while the second is based on Lemma (9). The third equivalence follows again from the Causal Markov and D-Faithfulness assumptions, while the last one is based on the definition of D-separation, which reduces to d-separation when conditioning on a set Z for which DET(Z) = Z. Using the result from Theorem 10, we can now introduce our strategy for dealing with faithfulness violations due to deterministic relations. First we rephrase a constraint-based algorithm in terms of d-separations and d-connections, which is usually a trivial change, as shown in Section 6. Then we can convert the problem of possibly unfaithful independences to the problem of possibly incomplete input. Specifically, we can derive a subset of sound d-separations and d-connections from independence test results as follows: Corollary 11 Let X, Y, W be disjoint (sets) of variables. Let DET(W ) be the variables determined by (a subset of) W (see Definition 3). Under the Causal Markov and D-Faithfulness assumptions, the following holds: Proof The first implication follows from the Causal Markov assumption, while the second follows from the Causal Markov and D-Faithfulness assumptions, and Theorem 10. Note that this procedure outputs d-separations only for a subset of independence test results, ignoring independences when X or Y ∈ DET(W ). The simple strategy in Corollary 11 can be applied to any constraint-based method, providing that it can deal with partial inputs, i.e., missing results for certain independence tests. Logic-based methods (e.g., Hyttinen et al., 2014;Magliacane et al., 2016) can be run out-of-the-box with partial inputs, while other standard algorithms like FCI (Zhang, 2008) would require possibly non-trivial extensions. Anytime FCI (Colombo et al., 2012) allows one to ignore (in)dependences above a certain order, but up to that order they are all required to be available, so that algorithm would also require possibly non-trivial extensions. Ancestral Causal Inference With Determinism (ACID) We implement the strategy in Corollary 11 in Ancestral Causal Inference with Determinism (ACID) as a determinism-tolerant extension of Ancestral Causal Inference (ACI), a recently introduced logic-based method (Magliacane et al., 2016). Before describing how ACID differs from ACI, we will briefly describe ACI itself. Ancestral Causal Inference (ACI) ACI reconstructs ancestral structures (combinations of "indirect" causal relations), also in the presence of latent variables and statistical errors. Ancestral structures are formally defined as: Definition 12 An ancestral structure is any relation on the observed variables that satisfies the non-strict partial order axioms: The underlying causal DAG induces a unique ancestral structure on the observed variables: the transitive closure of the direct causal relations (directed edges) in the DAG. ACI encodes the ancestral structure definition and five other causal reasoning rules: Lemma 13 For X, Y , Z, U , W disjoint (sets of) variables: These rules are shown to be sound assuming the Causal Markov and Causal Faithfulness assumptions. Causal discovery is then reformulated as an optimization problem where a loss function is optimized over possible ancestral structures. Given a list of weighted inputs, e.g. a set of conditional independences weighted by their confidence, the loss function sums the weights of all the inputs that are violated in a candidate ancestral structure. In addition, ACI provides a method for scoring causal predictions, which roughly approximates their marginal probability. ACID Out-of-the-box ACI is not able to deal with the faithfulness violations due to deterministic relations, and thus cannot be used for JCI. Therefore, we propose Ancestral Causal Inference with Determinism (ACID), which extends ACI following the strategy discussed in Section 5. We reformulate the logical rules of ACI in terms of d-separation, completely decoupling them from any assumption on the relation between (in)dependences and d-separations/connections, e.g., Causal Faithfulness. These new rules, that we call the ACID rules, are almost identical to the original ACI rules, except that the independences ⊥ ⊥ are substituted by d-separations ⊥ d , and the dependences ⊥ ⊥ by d-connections ⊥ d , as we show in the following: Lemma 14 For X, Y , Z, U , W disjoint (sets of) variables: Proof The proofs of these rules are slight modifications of the proofs of the ACI rules. For completeness, we provide them in the Appendix. So far, this only changes the interpretation of the implemented rules, but no change of the code is required. What changes are the inputs: only the sound d-separations and d-connections that can be derived with Corollary 11 are used as inputs for ACID. Similarly to other logic-based methods, the ACID rules are sound also with partial inputs (i.e. when some d-separation information may not be available). On the other hand, using partial inputs may reduce the completeness of causal discovery. We consider this a minor issue, since our focus is on prediction accuracy, and ACI is already known not to be complete in the general case, but has nevertheless been shown to obtain state-of-the-art accuracies (Magliacane et al., 2016). ACID-JCI To improve the identifiability and accuracy of the predictions, we also add as background knowledge a series of logical rules describing the causal structure of the regime and intervention variables that apply in the JCI setting: Lemma 15 Under the JCI assumptions, for any set of variables W : 8. ∀i ∈ I, ∀j, k ∈ X s.t. j = k and X j , X k ∈ W : Proof The propositions follow directly from the JCI assumptions and background knowledge: 1. R causes the intervention variables directly; 2. R cannot cause system variables directly, but only through intervention variables; 3. The intervention variables suffice to block any path between R and any other variable; 4. System variables cannot cause any I i or R; 5. There are no confounders between R and the system variables; 6. There are no possible confounders between the intervention variables and system variables other than R; Adding this background knowledge provides a simple means to ruling out several spurious candidate causal structures that do not satisfy the JCI modeling assumptions. The integration of such complex knowledge is one of the main advantages of logic-based methods. We will refer to the combination of ACID with these rules as ACID-JCI. If some of the intervention targets are known, we can also use this information as background knowledge. For example, if we know that the inhibitor I Akt−Inh targets the protein Akt, we can simply add the background knowledge I Akt−Inh Akt. Given the flexibility of logic-based approaches, we can also include more complex cases. For example, for two mutually exclusive targets A and B for intervention I 1 , we can add the background knowledge: I 1 A ∨ I 1 B, I 1 A ⇐⇒ I 1 B and I 1 B ⇐⇒ I 1 A. Evaluation We evaluate ACID on simulated data in the JCI setting. The simulator builds up on a simulator used in related work (Hyttinen et al., 2014;Magliacane et al., 2016) and implements soft interventions on unknown targets. For each combination of the number of system variables p and interventions i, we generate randomly 1000 linear acyclic models with latent variables and Gaussian noise, and simulate soft interventions on random targets. We then sample N = 500 data points for each model, randomly distributed between the i experimental datasets and the observational dataset, perform independence tests and weight the (in)dependence statements using the weighting schemes described by Magliacane et al. (2016). In our setting, we evaluate the causal discovery methods applied to datasets with unknown-target soft interventions. Other existing constraint-based methods (e.g., Hyttinen et al., 2014;Triantafillou and Tsamardinos, 2015) do not apply to this setting as they assume perfect interventions with known targets. Moreover, we wish to evaluate the net effect of using JCI with respect to merging separately learnt causal structures, while factoring out the effect due to different algorithms. Therefore, we choose to compare the ancestral structure predicted by ACID-JCI with a naive baseline based on ACI, "Merged ACI". In this baseline we merge ancestral structures learnt on each dataset separately with ACI by averaging the scores of the causal predictions over the datasets. As inputs to both algorithms we provide the same weighted independence test results (up to maximum In Figures 6 and 7 we report the precision and recall (PR) curves for predicting ancestral relations ("indirect" causal relations) and non-ancestral relations (the absence of such a causal relation) for different settings of p system variables and i interventions. We can see from these figures that ACID-JCI improves significantly on the accuracy of the predictions with respect to the baseline. As expected, the more interventional datasets are available, the better the accuracy for ACID-JCI, as we can see in the case of p = 4, i = 1 vs. p = 4, i = 3 in Figure 6. When there are only few datasets, e.g., i = 1, the performance of both methods improves as the number of system variables p increases, and the gap between the method gets smaller, as we can see in the case of p = 4, i = 1 and p = 6, i = 1 in Figure 6. On the other hand, when there are only few variables, e.g., p = 3 in Figure 7, ACID-JCI is able to predict correctly several ancestral relations that are not identifiable otherwise. In particular, in the case of p = 3 there are no predicted ancestral relations for standard methods like "Merged ACI", which explains why the PR curve for ancestral relations in Figure 7(a) starts at 0.5 recall. Instead, ACID-JCI is able to accurately predict these relations, illustrating that the Joint Causal Inference framework not only leads to statistical advantages but also enlarges the set of identifiable ancestral relations compared to methods that deal with each dataset separately. Conclusions and discussion In this paper, we presented Joint Causal Inference (JCI), a powerful formulation of causal discovery over multiple datasets that has been unexploited so far by constraint-based methods. Current constraint-based methods cannot be applied out-of-the-box to JCI because of faitfulness violations, so we proposed a simple strategy for dealing with this type of faithfulness violations. We implemented this strategy in ACID, a determinism-tolerant extension of a recently proposed causal discovery method, and applied ACID to JCI, showing its benefits in an evaluation on simulated data. A limitation of JCI is that the assumption of a unique underlying causal DAG precludes certain types of interventions. There are several techniques to extend our formulation of the problem to perfect interventions, or other interventions that induce new independences. For example, given an observational dataset, we could identify the datasets with perfect interventions by noticing the additional independences, perform causal inference on each of them separately and merge the predictions in a similar way to the methods presented by Hyttinen et al. (2014) and by Triantafillou and Tsamardinos (2015). In future work, we plan to investigate these techniques, as well as techniques to include fat-hand interventions that induce new dependences between the intervention targets. Moreover, we plan to investigate other possible strategies or extensions to existing algorithms for dealing with faithfulness violations due to deterministic relations. Finally, although very accurate and flexible, logic-based methods as HEJ (Hyttinen et al., 2014) and ACI (Magliacane et al., 2016) are limited in the number of possible variables they can handle. JCI introduces additional variables, reducing their scalability even more. We plan to investigate improvements to the execution times of methods like ACID. 4. If in addition, X ⊥ d Z \| W ∪ U , then U must be a noncollider on the subpath X . . . Z. Therefore, X ⊥ d Y \| W ∪ U . 5. Assume that Z ⊥ d X \| W and Z ⊥ d Y \| W . Then there must be paths π between Z and X and ρ between Z and Y in G such that each noncollider is not in W and each collider is ancestor of W . Let U be the node on π closest to X that is also on ρ (this could be Z). Then we have a path X · · · U · · · Y such that each collider (except U ) is ancestor of W and each noncollider (except U ) is not in W . This path must be blocked given W as X ⊥ d Y \| W . If U would be a noncollider on this path, it would need to be in W in order to block it; however, it must then also be a noncollider on π or ρ and hence cannot be in W . Therefore, U must be a collider on this path and cannot be ancestor of W . We have to show that U is ancestor of Z. If U were a collider on π or ρ, it would be ancestor of W , a contradiction. Hence U must have an outgoing arrow pointing towards Z on π and ρ. If we encounter a collider following the directed edges, we get a contradiction, as that collider, and hence U , would be ancestor of W . Hence U is ancestor of Z, and therefore, X ⊥ d Y \| W ∪ Z.	2017-03-09T12:43:04.000Z	2016-11-30T00:00:00.000	{ "year": 2016, "sha1": "157f96655484cace389888c603616be31c5af34a", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "157f96655484cace389888c603616be31c5af34a", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Computer Science", "Mathematics" ] }
4965495	pes2o/s2orc	v3-fos-license	Single photon triggered dianion formation in TCNQ and F4TCNQ crystals Excited state dynamics in two strong organic electron acceptor systems, TCNQ and F4TCNQ single crystals, was studied. After absorption of a single photon, dianions are formed in both crystals on ultrashort timescale: TCNQ τ < 50 fs, F4TCNQ τ = 4 ps. By use of transient absorption spectroscopy, we demonstrate that the dianion formation in F4TCNQ is mediated by the radical anion precursor which is described by a two-step model. Our measurements show the phenomenon that in this quinoid acceptor crystals in the absence of additional donor molecule, it is possible to resolve the two step formation of a doubly charged anion upon absorption of a single low energy photon (2.6 eV). -Tetracyanoquinodimethane (TCNQ) is a strong organic electron acceptor, with a high electron affinity of 2.8 eV 1,2 -3.38 eV 3 . Therefore, TCNQ has been used for synthesis of large number of charge transfer compounds that have been widely explored as building blocks of molecular electronics 4,5 , nonlinear optics 6 , and organic semiconductors 7,8 . A tetrafluorinated derivative of TCNQ, F 4 TCNQ (2,3,5,6-Tetrafluoro-7,7,8,8-tetracy anoquinodimethane), achieves an even higher electron affinity of 5.2 eV due to the presence of the four fluorine atoms 9 . The fluoride substitutes change the electrostatic potential but keep the same molecular geometry as that of TCNQ 10 . The charge transfer compounds of TCNQ and F 4 TCNQ show unusual electric and magnetic properties, ranging from metallic conductors, semiconductors to insulators [11][12][13] . After electron uptake from donor molecules, TCNQ and F 4 TCNQ are reduced to negative charged anions. These anions also classified as radical anions, TCNQ •− and F 4 TCNQ •− , are among the very few anions which possess bound excited electronic states 14,15 . In crystals these bounded excitons can decay back to neutral molecules, dissociate to conducting electrons and holes or undergo a subsequent chemical reaction forming dianions from anions by further electron reduction after applying voltage 16 or in the gas phase by collision with the sodium vapor 17 . It is noteworthy that in order to form the anion or dianion of TCNQ or F 4 TCNQ, other molecules are necessarily used as donors. These additional donors may be other molecules as tetrathiafulvalene (TTF) in charge transfer compound TTF-TCNQ or, as shown in this paper, neutral TCNQ or F 4 TCNQ. The last reaction is highly unusual. In the TCNQ and F 4 TCNQ crystals only few molecules are excited from the ground state and neighboring molecules may serve as electron donors leading to the formation of a radical ion pair according to reaction 1 when exciting the TCNQ crystal: The driving force for this reaction is largely determined by the photon energy absorbed by TCNQ molecules in the crystal lattice. However, the anion of TCNQ − is chemically unstable. It may recombine back to two neutral TCNQ molecules in their electronic ground states through reaction with a cation TCNQ + emitting phonon or it can be further reduced by electron uptake from an adjacent neutral TCNQ molecule according to the following reaction: Scientific RepoRts \| 6:28510 \| DOI: 10.1038/srep28510 Existence of all five species, TCNQ, TCNQ * , TCNQ − , TCNQ 2− , and TCNQ + taking part in the above mechanism has been experimentally observed. Therefore, it is of high interest to explore this complex reaction scheme in a pure condensed phase containing only the strong quinone acceptors. This two-step process is triggered by a single-photon absorption with very low photon energy (2.6 eV). In this work, we present the time-resolved spectroscopic study in TCNQ and F 4 TCNQ single crystals using femtosecond transient absorption (TA) and time-resolved fluorescence spectroscopy. Excited state dynamics reveals that after absorption of a single photon, dianions are formed in both crystals on an ultrashort timescale (< 4 ps). We also demonstrated that the dianion formation is mediated by the radical anion precursor and described by a two-step model. Results Steady-state spectra and time-resolved fluorescence. The steady-state absorption spectra of the neutral molecule, radical anion, dianion, and single crystals of TCNQ and F 4 TCNQ are shown in Fig. 1. Absorption maxima of neutral TCNQ and F 4 TCNQ in acetonitrile locate at 393 nm which correspond to transition S 0 → S 1 . Due to the unpaired single electron in TCNQ and F 4 TCNQ anions, their electronic structure contains doublet states (labeled as D). F 4 TCNQ •− and TCNQ •− have similar absorption spectra: two main absorption bands, one at 410 nm corresponding to the D 0 → D 2 transition, and the other at 600-900 nm with two local maxima of 754 and 856 nm, due to the D 0 → D 1 transition 17,18 . In ref. 17, Panja et al. studied absorption and lifetime of the F 4 TCNQ 2− in the gas and solution. The major absorption band is located at 335 nm. Additional broad absorption in the 400-700 nm region is also observed both in the gas and solution. Both spectra can be described by the sum of two Gaussians (481 and 567 nm) 17 . However, in solution, the oscillator strength of 567 nm band is much stronger than that in the gas phase. In our recent study 19 , we have characterized the spectroscopic properties of TCNQ in acetonitrile. The fluorescence recorded at room temperature is very weak (quantum yield φ < 10 −4 ), with ultrafast decay time (800 fs). In the case of single crystals, both TCNQ and F 4 TCNQ crystals also do not show detectable steady-state fluorescence at room temperature. Their fluorescence lifetimes were measured by fluorescence upconversion technique (Fig. S2). There are two decay components in both crystals. In F 4 TCNQ, τ 1 = 0.3-0.8 ps with amplitude about 70%, τ 2 ranges from 4.5 to 12 ps with amplitude about 30% (Supplementary Table S1 and Fig. S2b). The ultrafast fluorescence indicates that there are efficient quenching channels as discussed below. Transient absorption spectra. The transient absorption spectra of TCNQ and F 4 TCNQ crystals were measured under excitation at the absorption edge. The TA spectra and kinetics of TCNQ and F 4 TCNQ crystals In TCNQ crystal, this TA signal shows an instantanous rise, followed by an ultrafast decay with lifetime of 150 fs (Fig. 2b) which is comparable to the instrument response function (IRF) of the pump probe setup. This ultrafast process is due to nondegenerate two photon absorption (ND-TPA) 20 . ND-TPA is a process when one photon from the pump and one photon from probe beam are simultaneously absorbed by the sample, and thus the molecule appears in two-photon excited state. This process was observed only at λ exc = 475 nm, i.e., in the absence of linear absorption. In the presence of strong one-photon absorption (excitation at 250 or 350 nm), this ultrafast process disappears (Supplementry Figs S5 and S6), i.e. no ND-TPA. In F 4 TCNQ crystal, the 557 nm TA band shows a slow rise (4.3 ps). Moreover, there is an additional weak TA band between 650 and 850 nm (Fig. 3a) which decays within 5 ps. This decay corresponds well with the rise of 557 nm TA band, indicating that one transient is transferred to the other. The origin of this transformation process is discussed below. Discussion In F 4 TCNQ crystal, the peak position of the weak TA band between 650 and 850 nm (Fig. 3a) corresponds well with the absorption of radical anion F 4 TCNQ •− (see Fig. 1b blue line). Therefore, we assign this band to the absorption of F 4 TCNQ •− . In Fig. 3a, for comparison, the absorption spectrum of F 4 TCNQ 2− in solution (Fig. 1b green line) is plotted together with our TA spectra. The location and shape of the 557 nm TA band are in excellent agreement with the absorption spectrum of F 4 TCNQ 2− . In order to estimate the lifetime of this long-lived TA band, we also plotted the TA spectra at negative delay time (−3 ps), which can be considered as 1 ms time delay due to our 1 kHz laser repetition rate. We can see that the TA spectra at > 550 nm show no visible decay within 4 ns but fully disappear within 1 ms, indicative of a lifetime being 4 ns < < τ 1 < 1 ms. However, in the probe range < 550 nm, there is visible TA signal at negative delays. By a first order exponential fit, we can estimate this lifetime to be about 0.6 ms. This value also corresponds well with the reported F 4 TCNQ 2− lifetime of 0.7 ms (30%) 17 . The other two longer lifetimes of F 4 TCNQ 2− 15 ms (60%) and 1s (3%) reported in ref. 17 cannot be observed in our TA measurements due to the limitation of the 1 kHz laser. Therefore, we assign the TA band at 557 nm to the absorption of F 4 TCNQ 2− . As discussed above, the TA band between 650 and 850 nm (Fig. 3a) is due to the radical anion F 4 TCNQ •− , and 557 nm TA band is due to F 4 TCNQ 2− . From the kinetics of both F 4 TCNQ •− and F 4 TCNQ 2− (Fig. 3b), we can see that the F 4 TCNQ •− is generated instantaneously (< 50 fs) and decays within 5 ps. Moreover, F 4 TCNQ 2− (557 nm We also measured the TA spectra of F 4 TCNQ under excitation at 250 and 350 nm, shown in Figs S3 and S4 (Supplementry Information). Compared with the TA spectra under 475 nm excitation, they show the same 557 nm TA band but no visible F 4 TCNQ •− absorption in 650-850 nm region. Moreover, in addition to the 4 ps rise time, an instantaneous rise component (< 50 fs) appears in the TA kinetics. This additional ultrafast formation time under excitation at higher photon energies suggests that there is a competitive channel of dianion formation, i.e. direct two-electron transfer from the highly excited F 4 TCNQ to the unexcited neighboring molecules. TCNQ can be considered as a analogue of F 4 -TCNQ, since their neutral and anion absorption spectra are very similar, which indicates the four fluoride substitudes do not remarkably distort the electronic structure. At time delays longer than 300 fs, the TA spectra of TCNQ crystal present a similar TA band at 557 nm, however, broader than that of F 4 TCNQ (Fig. 2a). This TA band is also long-lived without noticeable decay within 4 ns (Fig. 2b). Therefore, the 557 nm TA band in TCNQ crystal is also assigned to dianion (i.e., TCNQ 2− ). The TA kinetics of TCNQ 2− does not show any rise, indicative of an ultrafast dianion formation (< 50 fs). It is noteworthy that the anion absorption of TCNQ crystal is red shifted compared with TCNQ solution (Figs 1a and 2a) while the anion absorption from F 4 TCNQ crystal is similar to F 4 TCNQ solution (Figs 1b and 3a). We explain this due to the molecular packing difference in two crystals. In TCNQ crystal, adjacent TCNQ molecules pack in a cofacial manner (Fig. S1, Supplemental Material) which facilitates electron delocalization between neighbor TCNQs, resulting in a red shifted anion absorption. However, in F 4 TCNQ crystal, molecules pack in a fashion (Fig. S1, Supplemental Material) that the spatial overlap is poor between adjacent molecules. Therefore, the anion absorption in F 4 TCNQ crystal is less affected. A similar example can be found for NDI molecule which is also a well-known electron acceptor. In a cofacial stacking NDI dimer, the anion is more red shifted relative to the monomer anion 21 . However, in a triangular NDI prism where the cofacial overlap is much worse, the anion absorption spectrum is very similar to the NDI monomer solution 22 . The experimentally observed long absorption tails (Figs 2a and 3a) indicate the presence of a charge transfer transition in both TCNQ and F 4 TCNQ crystals, therefore, we explain the dianion formation mechanisms in TCNQ and F 4 TCNQ crystals with a two-step model: the initial excitation to a charge transfer excitation forms a cation and an anion, followed by a second step that is an electron transfer between the anion and an adjacent neutral molecule, forming a dianion. The dianion generation reaction scheme after band edge excitation of F 4 TCNQ is depicted in Fig. 4. Qualitativelly, the model and reactions for both TCNQ and F 4 TCNQ are similar. Within 50 fs after photoexcitation, charge transfer excitation results in a cation F 4 TCNQ + and an anion F 4 TCNQ •− ; then this F 4 TCNQ •− within 4 ps adopts another electron from an adjacent neighbor F 4 TCNQ to form the double charged anion, i.e., dianion F 4 TCNQ 2− . However, in TCNQ crystal, the second step is ultrafast, resulting in the absence of TCNQ anion intermediate in the TA spectra. The molecular and electronic structures of TCNQ and F 4 TCNQ are very similar; the only apparent difference between TCNQ and F 4 TCNQ crystals is their crystalline structure: F 4 TCNQ crystal is orthorhombic while TCNQ crystal belongs to the monoclinic system, as shown in Supplementry Fig. S1. We can see that in TCNQ crystal, along the stacking direction, the neighbor molecules are well parallel to each other, i.e., they are in a stacking geometry with strong charge density overlap, resulting in a larger electronic coupling H el . However, in F 4 TCNQ, the molecular stacking is not as good as that of TCNQ, which will induce a poorer charge density overlap thus a smaller electronic coupling H el . Therefore, difference in rise times confirms that the dianion formation originates from the intermolecular interaction. It is in line with our conclusion that dianions are formed via electron transfer between neighboring molecules. To sum up, in two strong organic electron acceptors TCNQ and F 4 TCNQ crystals, for the first time we observed a formation of single-and double-charged anions due to charge transfer between excited acceptor molecule and neutral acceptor molecules after absorption of one photon. In both crystals, after photon absorption, the anions are formed almost instantenously (within 50 fs). Next step includes formation of dianions by another electron uptake from the neighbor molecules. The dianion formation time is structure dependent: in the well packed monoclinic TCNQ crystal, TCNQ 2− forms within 50 fs. However, in the orthorhombic F 4 TCNQ crystal where the adjacent molecules are not well packed, F 4 TCNQ 2− forms much slower: within 4 ps. In the presence of different donor molecules, TCNQ and F 4 TCNQ may form charge transfer compounds with multiple stoichiometry. Such compounds have already been observed for perylene-TCNQ (1:1, 1:2, 1:3) 23,24 but still not in perylene-F 4 TCNQ. Our work shows that in the absence of additional donor molecule, in the well packed strong acceptor systems photoexcitation leads to formation of doubly charged ions. Absorption and fluorescence. Steady-state absorption spectra were obtained by a double beam UV-Vis spectrophotometer (Cary 100Bio, Varian). The fluorescence upconversion measurements were carried out by a system FOG 100 (CDP Systems). The second harmonic of a Titanium sapphire laser (Chameleon, Coherent Inc.) at 400 nm (100 fs, 80 MHz) was used as the excitation source. The fluorescence (430-650 nm) was collected by parabolic mirrors and focused onto a 0.5 mm BBO crystal (cut angle 38°, type I nonlinear interaction) together with the fundamental radiation (800 nm) to generate the sum-frequency radiation. The resulting radiation (280-359 nm) was detected by a photomultiplier based photon counting electronics after passing through a double monochromator (CDP2022D). Femtosecond laser spectroscopy. The transient absorption (TA) spectra were measured by the optical femtosecond pump probe technique. The output of titanium-sapphire (Legend Elite, Coherent) regenerative amplifier seeded by the oscillator (Micra, Coherent) was used as a pulse laser source. The output laser beam was with wavelength 800 nm, pulse width 65 fs, pulse repetition rate 1 kHz, and average power 3.5 W. This fundamental radiation was converted to 250-475 nm by use of optical parametric amplifier (Topas, Light Conversion); it was used as the pump beam. White light continuum generated in a sapphire plate was used as the probe beam 27 . The diameter of pump and probe beams are about 50 μ m at focus point. The intensity ratio between pump and probe is about 20:1. The pump polarization is fixed at magic angle (54.7 degree) relative to the probe polarization (horizontal). The crystals were fixed at the edge of glass slides using double-sided adhesive with their long axes (b axes) parallel to the probe polarization (horizontal direction). The absorbance change ∆A is calculated by measuring the transmitted probe intensity in the presence (I pump_on ) and in the absence (I pump_off ) of the pump beam:	2017-11-13T22:40:03.639Z	2016-06-27T00:00:00.000	{ "year": 2016, "sha1": "747479477f3cfc420b354bd752cda5ee3686858f", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/srep28510.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "cf564d0102bc1ead14bc591f181dc54458e89d79", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Medicine", "Materials Science" ] }
245884522	pes2o/s2orc	v3-fos-license	New business models at distribution grids: a stakeholder consultation In this paper, four business models are presented and discussed from the perspective of different stakeholders or actors in the ecosystem, either because they are involved in their implementation or affected by them. This discussion derives from interviews conducted with 31 representatives from industrial consumers, energy regulators, policymakers, distributed energy resource (DER) owners, TSOs, DSOs, retailers, aggregators, ESCOs, and data service providers. The consultation covered experts from four European countries, namely Portugal, Sweden, Slovenia, and Spain. The business models under consultation included the (i) use of flexibility by DSOs, (ii) aggregation through the Virtual Power Plant (VPP) concept, (iii) consumers (industrial and residential) energy usage optimization and flexibility provision, and (iv) the concept of data service provision through data platform. We conclude by identifying the most relevant drivers and barriers identified for each business model. Barriers are subdivided for each business model according to three categories, namely (i) regulatory, (ii) technological, and (iii) organizational or behavioural. Introduction The deployment of smart grids, the empowerment of consumers and energy use decarbonization is not only changing the way electricity is produced, distributed, and consumed, but is also opening the possibility for new business models (BMs) [1], [2]. In this paper, we analyse four new BMs at distribution grids. BMs are here understood as a set of strategies chosen by a main actor in order to generate economic benefit. These strategies can combine multiple business plans, and the economic benefits can be generated by different sources of revenue streams and/or cost reductions. The BMs under consultation included the (i) use of flexibility by DSOs, (ii) aggregation through the Virtual Power Plant (VPP) concept and provision of service to both DSOs and TSOs, (iii) consumers (industrial and residential) energy usage optimization and flexibility provision, and (iv) the concept of data service provision. The latter could be considered the most innovative BM, in which the main actor, new or incumbent (e.g. retailers, ESCOs) would provide data services (e.g. forecasts, energy optimization) based on the metering and other data that could be made available by data hubs. A complete description and discussion of each BM is found in [3]. This paper leverages on the work developed within the H2020 InteGrid project. Additional information on the definition of BMs and the stakeholder consultation process and results can be found in [4] and [5], respectively. Methodology For this consultation, relevant stakeholder for each BM were identified. For that purpose, a power/attention matrix is used, based on the methodology proposed in [6]. In this paper, authors interviewed representatives from the key actors in four different countries, namely Portugal, Slovenia, Spain and Sweden, following an in-depth personal interviewing methodology. Respondents were asked to discuss the perceived drivers and barriers, as well as other financial, regulatory or social feasibility issues related to these new business models. In total, 31 personal interviews took place. To preserve anonymity, the role of the actor is reported, rather than personal details that could make the informant identifiable (Table 1). In addition, citation attribution is not made per nationality if this might make the informant identifiable. Interviews followed a semi-structured approach. The analysis of the interviews led to the identification of drivers and barriers for each BM. Barriers were classified into regulatory, technological and organizational/ behavioural. Additionally, the "level of agreement" was also observed. A high level of agreement means that most or all participants agree with the driver/barrier, while a medium level of agreement means that two or more participants mentioned/agreed on the same barrier. However, most drivers/barriers were mentioned by only one respondent, thus no level of agreement can be inferred. DSO procuring local flexibility for grid management This BM is defined as the procurement of local flexibility from distributed energy resources (DER) by the DSO (main actor) as a means to help manage the grid (e.g. solving local congestions or voltage problems). By doing so, the DSO could potentially benefit from reinforcement deferral or even avoidance. DSOs agree that procuring flexibility services will be part of their future, as a means to defer or avoid network reinforcements. However, they believe that regulation does not provide the certainty or incentives DSOs need to change the way they operate their networks. Firstly, the necessary economic incentives to promote the use of flexibility are missing. Secondly, they currently bear the full risk in case flexibility providers fail to provide the service, leading to grid problems and/or interruptions. Lastly, on the institutional realm, DSOs mentioned that they may face some internal resistance to adopt this BM. Regulators generally agree that promoting the use of flexibility by DSOs is necessary if DERs keep increasing as expected. They mentioned that the current CAPEXoriented regulation is an important barrier for DSOs, but did not have a clear view of where future regulation should go. Interviewees also mentioned the lack of local flexibility mechanisms as an important barrier. Some participants mentioned public consultations and pilots as a way to overcome this, while one interviewee remarked the importance to coordinate local flexibility procurement with tariff schemes. TSOs expressed their concerns about the inefficiencies of local mechanisms, especially market-based ones, due to their lack of liquidity. From the grid operation standpoint, TSOs mentioned that congestions in the grid should be the exception, and not the norm. Additionally, they mentioned that forecasting will become more difficult for the TSO, and that TSOs and DSOs will have to share the responsibility of the security of supply. Other stakeholders also commented on this BM. For instance, one DER owner deemed it difficult to provide the service to the DSO as they have no visibility over where and when flexibility will be needed, and the absence of clear pricing schemes. One policy maker mentioned that incentives are missing for both the DSO and the consumers. Retailers and ESCOs stated that additional hardware and communication would be necessary from their side. Likewise, aggregators were concerned about the lack of harmonization of local flexibility products and procedures across Europe, jeopardizing the scalability potential of aggregation tools and operations. Data services and platforms In this BM, a new (or incumbent) agent, here referred as a data service provider (DSP; main actor), uses metering data stored in central data hubs, if authorized to do so, to provide data services (e.g. forecasting, energy management, portfolio optimization for aggregators), either in B2B or B2C modalities. This BM captured different stakeholders' attention, including DSOs, industrial consumers, policy makers, retailers/ESCOs, aggregators, regulators, and a possible DSP. They emphasized the innovative approach that this BM implies. Paper 0582 Whilst they all agree that data platforms will necessarily develop and some stakeholders believe this will be useful, some stakeholders are sceptical of the value data platforms will create. The most relevant barrier, on which there is a general consensus, is the difficulties in accessing the data as required to provide data-driven energy services. The first reason may be the refusal or lack of interest of many end consumers to share their data. This can lead to a vicious circle in which consumers are not offered value due to the limited access to data, and consumers not accepting to sharing the data due to the low perceived value. As a possible solution, one DSP believed that some early consumers, who allow for using their data, could be offered cheaper retail deals in exchange, progressively leading to a change in behaviour. Nevertheless, another reason, which draws a very high level of agreement across stakeholders, is that data protection regulation seriously limits innovative data service opportunities. Besides the data access problem, interviewees identified two other important barriers. Firstly, a potential DSP mentioned that the EU is unlikely to harmonize the data access and procedures. That means that the DSP will have to create different solutions in different countries, which constitutes a major barrier, considering that data service provision will be a low-margin business and needs to be scalable to be profitable. Another risk for this BM identified by one stakeholder is that IT companies are already gathering electricity-related data (or inferring this data) outside the metering infrastructure. That means that these companies could soon be able to offer very similar services in a much more dynamic business model, as they are less bounded by energy regulations. Consumers reducing electricity bill and providing flexibility This BM is defined as the consumers (main actor), both residential and industrial, engaging in enhanced electricity management, adoption of distributed generation (DG) and possibly the provision of grid services, aiming at reducing the overall electricity cost. Several industrial consumers and one residential consumers' association participated in a discussion regarding this BM, as they are the main actors in this business model. Nevertheless, DER owners, policy makers, retailers and ESCOs, aggregators, regulators and TSOs, also provided their viewpoints. From the industrial consumer's standpoint, the interviews showed mixed opinions regarding the interest and possibilities for this BM. Some industries, usually the large and energy-intensive ones, are already very advanced in terms of internal energy management, whereas smaller and less energy-intensive industries are less concerned about optimizing energy usage (low benefits, or lack of resources to do so). These stakeholders mentioned high fixed regulated costs as a reason for the lack of interest in this BM. On the provision of flexibility to grid operators, the industrial consumers also see different barriers. Some interviewees said they would participate in service markets, and some already do it (e.g. in tertiary reserve). However, others mentioned that they would not participate, as this would imply them changing their production schedules for a small benefit. Additionally, some industrials mentioned that they could not change their consumption to comply with balancing products requirements (traditionally tailored to centralized generators). The residential consumer's association mentioned three main barriers, especially about flexibility provision: i) low price elasticity of consumers, ii) difficulty in understanding electricity markets, and iii) general mistrust in electricity companies. Therefore, consumers would be Paper 0582 less willing to give away the control of their consumption for what they anticipate is a reduced economic benefit. Other stakeholders also expressed their opinions and concerns. The DG owners, for instance, highlighted that energy and service markets (e.g. balancing) are not completely open yet for DER participation. It was also mentioned the need for enhanced infrastructure (communication) for these services to be provided. Aggregators mentioned that individual consumers are less aware or motivated regarding new possibilities, but that it could be more attractive for community users. However, one aggregator sees regulatory uncertainty over energy communities as a barrier. Regulators also commented extensively on this BM, particularly on the need to re-design electricity tariffs to promote an efficient behaviour from end-users. Interviewees recognized that current tariff structures do not promote flexibility, but several mentioned different barriers, such as the lack of sufficient historical metering data to support tariff design changes due to the recent smart metering deployment, fear of causing unintended consequences or complaints from end-users, or even the reluctance from some incumbents or policy-makers to these changes. Additionally, regulators identified aspects already mentioned before, such as the low interest from the consumer's side to adopt advanced tariff schemes due to the small benefit perceived. Finally, TSOs also expressed their expectations and concerns. They recognize that demand response (DR) participation will be very beneficial to overall efficiency of balancing markets. However, they expressed the possible internal resistance in TSOs due to doubts about the ability of DR to participate in complex and sensitive services for the system, such as fast frequency control (e.g. aFRR, FCR products). Interviewees identified technical requirements about observability, prequalification and product definition as barriers too. Flexibility provision through aggregation In this BM, the aggregator (main actor) provides services to grid operators (DSOs and TSOs) by aggregating different types of DER, exploring the concept and possibilities of the VPP. This business model explores in particular the VPP in two forms, namely the commercial VPP (cVPP), that provides services to the TSO (e.g. balancing markets), and the technical VPP (tVPP), which provides local flexibility to the DSO. This BM was mostly discussed with an independent aggregators and retailers (potential aggregators), the main actors in this business model, as well as regulators and TSOs. The independent aggregator interviewed acknowledges that service provision to DSOs and TSOs will be a viable business in the future, but not in the short term. They mention the wide variation of national market rules and designs as an important barrier, which limits the potential scalability. In other words, lack of standardization of market access interfaces is also a problem. Finally, they mention the need for real-time data to provide certain services (e.g. balancing). Regulators see different barriers for the cVPP and the tVPP concepts. Regarding the former, they mention that balancing products are still not completely adequate for DR participation. Additionally, that the framework for independent aggregators is underdeveloped, and that revenues from balancing markets may be limited. Regarding the tVPP, the barriers identified by regulators are similar to the ones for the first BM (local flexibility procurement by DSOs). Firstly, the revenue regulation for DSOs is still CAPEX-based, and, secondly, local flexibility mechanisms are not in place yet, limiting the amount of flexibility that DSOs are willing/can procure. Paper 0582 The TSOs also mentioned the lack of role definitions for VPPs, balance responsible parties (BRPs) and balancing service providers (BSPs) as a barrier. In addition, they mention that VPPs may make the forecasting process more difficult, especially if these VPPs are large and resources are scattered across different regions.	2022-01-13T16:30:52.317Z	2021-01-01T00:00:00.000	{ "year": 2021, "sha1": "f5aecd738567f1abba9c6bece0343693ee8f8777", "oa_license": "CCBY", "oa_url": "https://zenodo.org/record/5608337/files/CIRED21_Stakeholders_v2.1.pdf", "oa_status": "GREEN", "pdf_src": "IEEE", "pdf_hash": "5bb51acfe81b6464441f592b610d9db9995b1d00", "s2fieldsofstudy": [], "extfieldsofstudy": [] }
21370540	pes2o/s2orc	v3-fos-license	Comprehensive functional genomics using Caenorhabditis elegans as a model organism We have been working on functional genomics using C. elegans as a model organism. We first used cell-type specific markers and preexisting mutants to investigate how genotype-phenotype causal relationships are regulated. With the aid of transgenic methods, we analyzed various biological processes in C. elegans. We have developed efficient methods to isolate gene knockout strains. Thousands of strains isolated this way are used by many researchers and have revealed many biological mechanisms. We have also developed methods to examine the functions of genes in a comprehensive manner by integrating transgenes into chromosomes, designing conditional knockouts, and creating balancers for lethal mutations. A combination of these biological resources and techniques will be useful to understand the functions of genes in C. elegans, which has many genes that are orthologous to those of higher organisms including humans. Introduction Medical and life science research studies involve many disciplinary approaches, including genetics. Strategies using genetics are characterized by tight causal relationships between genotypes and phenotypes, although these are often indirect. Many of the earlier findings of genetic studies were descriptive and did not reveal cellular and molecular mechanisms. Genetic studies of human diseases are also based on the causal relationships between some genetic variations and symptoms or disease susceptibility. Importantly, the GWAS (genome-wide association study) and other new technologies such as highthroughput genome sequencing are now able to handle an enormous number of genes and, thus, are rapidly changing not only medical researches, but also clinical practice (see ref. 1-3 and cancer moonshot. Bethesda, MD: National Cancer Institute (https:// www.cancer.gov/research/key-initiatives/moonshotcancer-initiative)). However, in spite of the huge volume of information on the relationships between genetic variations and diseases, an increased understanding of individual gene function and interactions among genes would help us to understand their influence on the outcome. Genetics is also useful for investigating cellular and molecular mechanisms and genetic pathways. To fully understand the cellular and molecular mechanisms, it is useful to know the functions of genes and their products in relation to phenotypes found in experimental organisms and genetic epistatic analyses. To facilitate the acquisition of such experimental evidence, we have been using the nematode Caenorhabditis elegans (C. elegans), a small model organism. C. elegans has a large number of genes that are homologous to human genes and therefore, studies of gene function in C. elegans can help researchers understand the functions of human genes and their products. Moreover, C. elegans was the first model animal to have its whole genome sequenced. 4) The availability of whole-genome sequences for many organisms, including both experimental animals and humans, has facilitated research to elucidate the functions of disease-related genes. Here, I describe how C. elegans can be used as a tool for gene function analyses, mainly focusing on gene knockout, RNA interference (double-stranded RNA-mediated gene silencing), and transgenics, with which many C. elegans researchers are working. We have been isolating deletion mutants of C. elegans and have distributed them to many laboratories worldwide through the National BioResource Project for the nematode (ref. 5, http://shigen.nig.ac.jp/ c.elegans/index.jsp). By now, the cumulative number of publications using our mutants has exceeded 2,000, and it is difficult to summarize the entire contents of those papers by citing all the important publications. Therefore, here I will review the development of our functional genomics approaches using C. elegans and explain how C. elegans can be used for functional genomics studies by reviewing some publications that are relevant to our own experiences. C. elegans as a model organism for genetic analyses To describe how genetic analyses can be used to solve biological questions using C. elegans, a schematic diagram is shown (Fig. 1). Explanations of technical terms are described in the legend of Fig. 1 and historical examples are described in the following text. Half a century ago, Dr. Sydney Brenner proposed the use of the nematode C. elegans as a model organism to understand development and behavior because of its simple structure and short life cycle, compared with those of the various other experimental model organisms. Researchers working on C. elegans have been trying to elucidate its biological processes as a whole. The best-known examples are cell-lineage description, morphological reconstitution of electron micrographs, and genome sequencing. 4),6)- 8) C. elegans is a small animal, only 1 mm long at the adult stage, and mainly proliferates as a hermaphrodite. The animals can be easily cultured on agar plates and fed with Escherichia coli. They can be stably stored for years in deep freezers or liquid nitrogen. The organism has only approximately 1,000 somatic cells, including 302 neurons and 7,000 synaptic connections that form neural networks. 6), 8) C. elegans worms grow from fertilized eggs to adulthood in only 3 days at 20°C. C. elegans has two sexes: hermaphrodite and male. Hermaphrodites have 5 pairs of autosomes and one pair of sex chromosomes (X), while males have only one sex chromosome and the same number of autosomes. Hermaphrodites have both oocytes and sperm, with which they self-fertilize to give rise to progeny, making it easy for researchers to obtain homozygous animals by culturing a single hermaphrodite. Males are found at a rate of approximately 1 in 500 selffertilized hermaphrodites because of the loss of one X chromosome during the meiosis. The genome of C. elegans, approximately 100 Mb in size, is much smaller than the human genome of 3 Gb. Thanks to the organism's small body and detailed basic descriptions with many experimental tools, researchers can use C. elegans as proliferative "test tubes". Brenner published a commemorative paper in 1973 in which he described hundreds of mutations mapped on 6 chromosomes (linkage groups) of C. elegans. 9) These mutations served as genetic landmarks in the following analyses by researchers. After having obtained the genetic map, researchers tried to isolate many mutants with abnormalities, including cell lineage, apoptosis, behavior, and morphogenesis etc. 10) Early efforts focused on the forward genetics, which involved isolating mutants with interesting phenotypes and then positionally cloning the causative genes. In the 1980s, it took researchers a long time to identify a gene that causes a given phenotype. Researchers added new genes to the genetic map, which eventually made it easier to identify other genes and sequence the genome. In 1992, the C. elegans Sequencing Consortium reported its progress on chromosome III, and in 1998, the consortium published a paper that described the whole C. elegans genome sequence and the various characteristics of the genome structure. 4),11) The genome analysis revealed that C. elegans has approximately 20,000 protein-coding genes and that they are homologous to protein-coding genes found to be addressed by genetical analyses are indicated by orange boxes, whereas methods to solve the questions are indicated by blue boxes. Representative questions are originated by multiple layered biological processes, multiple pathways for similar phenomena, structure-function relationships of genes and their products, and effects of environments. Multiple layered biological processes mean that organisms have multiple layered regulations of causal relationships with molecular-, cellular-, tissue-and whole body-levels. Also, organisms have spatio-temporal regulations. Complicated mechanisms are unraveled, for example, by examining transgenic analyses mentioned in the text. Multiple pathways for similar phenomena mean that organisms have often multiple independent causes for seemingly similar phenotypes and these are categorized and unraveled by crossing with other strains (as epistatic analyses) and isolation of enhancers and suppressors to identify genetic interactions. Effects of environments are important subjects because gene regulations are essential for adaptation to the environment an organism is exposed. Explanation of blue boxes are followings. The phrase "expression analyses by transgenic strains" indicates that transgenic strains can be generated by injecting plasmids into the germline of the animals. The phrase "in vitro plasmid construction" means that researchers can generate varieties of plasmids with current molecular biological methods. The phrase "reverse genetics, forward genetics and phenotyping" means that researchers can isolate new mutants only by sequences and only by the inspection of phenotypes of interest as "reverse genetics" and "forward genetics", respectively. The phrase "enhancer/suppressor screening with mutagenesis or RNAi" means that researchers screen for new mutants (or RNAi clones) which aggravate (enhancer) or attenuate (suppressor) the phenotypes of parent mutants. The phrase "crossing with other mutants, transgenic strains and phenotyping" means that researchers make double mutants etc. to see which phenotype is dominant, to dissect whether two causal genetic variations are interactive with each other. The phrase "mutations identified by whole genome sequencing" means that researchers can identify any types of mutations by analyzing the whole C. elegans genome using the next-generation sequencers and bioinformatics. The phrase "transgenic rescue with wild-type and mutant construct" means that researchers can microinject various plasmids to see whether the sequences can rescue the mutant phenotypes to understand whether some genetic variants retain wild-type function or not. The phrase "dominant effects with wild-type and mutant constructs" means that researchers can know by overexpression or ectopic expression of wild-type or mutant proteins in the wild-type animal background whether they have dominant phenotypes or not. The phrase "effects of biological, chemical and physical insults" means that researchers can expose worms to some biological (for example, bacterial infection), chemical (for example, toxic substances) and physical (for example, hot temperature) and see the effects by phenotypes. The phrase "expression analyses by transgenic strains and transcriptome, phenotyping" means that researchers can use transgenic reporters to see the potential changes of the expression patterns under some environmental changes in addition to the information about when and where the gene products may play roles. Also, researchers can use the worms to analyze RNA expression changes with microarray or RNA sequencing with a next-generation sequencer to find the expression changes under some environmental changes in addition to phenotypes. Functional genomics of Caenorhabditis elegans No. 8] in other organisms, including humans. In addition, many more non-coding RNA genes were found. During the genome sequencing process and after the release of the whole genome, researchers often used this information to achieve new insights into the biological processes. Interestingly, the C. elegans Sequencing Consortium joined the Human Genome Project, which expedited our present understanding of the genetic system and human diseases. 12) After a number of bioinformatics analyses, a paper was published in 2011 that described an OrthoList of the C. elegans genome: 7,663 protein-coding genes (approximately 38% of C. elegans protein-coding genes) are orthologous to human genes, which are likely to be derived from common ancestors and have similar functions. Worms have many more homologous genes that constitute gene families related to human genes, which share sequence similarities but may not have the same functions. 13) Thus, in spite of its simple structure, C. elegans has many molecular similarities to other higher eukaryotes and is useful for examining cellular and molecular functions in a multicellular system. With the wealth of information on genome structure, researchers began to focus more effort on reverse genetics (see below). Another important characteristic of the C. elegans research is the sharing of strains and databases. C. elegans researchers have been sharing published strains through the Caenorhabditis Genetics Center (CGC) for decades (http://cbs.umn.edu/ cgc/home). Researchers shared established data by constructing ACEDB (A C. elegans DataBase). 14) Earlier on, researchers shared the database by copying the data through the community. In 2000, the successor database was established and became available through a website as WormBase (http:// wormbase.org). It contains almost all published C. elegans research, so that anyone can search the accumulated data for information of interest. Through this standardization, researchers can avoid unnecessary trials and easily reproduce previous experiments to proceed to further analyses. By using C. elegans as a model organism, researchers can examine genotype-phenotype relationships in a reproducible manner by referring to prior knowledge, sharing strains among laboratories and examining a large number of clonal animals. Analyses of nerve cell development as a beginning I began my C. elegans research work in 1989 in the Chalfie laboratory at Columbia University. During my stay in the laboratory, we analyzed how gene expression and cell specification are regulated, focusing on touch receptor neurons. Touch receptor neurons, which are composed of 6 cells (two ALMs, AVM, PVM and two PLMs), had been analyzed by forward genetics, and important developmental and functional genes had been described. 15)-17) As an early phase of the project, we tried to develop a whole mount in situ hybridization method that allowed detection of gene expression at a single-cell resolution and was applicable to many genetic backgrounds without crossing strains to transgenic reporter strains such as lacZ fusion-expressing transgenic animals. We successfully detected the specific expression of two genes, ben-1 (benzimidazole resistant) and mec-7 (mechanosensory abnormality), both of which are O-tubulin, a major microtubule protein that forms a dimer with ,-tubulin. Our efforts to understand gene functions via genotype-phenotype causal relationships began with an experiment on developmental neurobiology, which was published in 1993. 18) In the course of the experiments, we used a combination of touch receptor neuron-specific markers including mec-7 and approximately 50 preexisting mutants that had shown some developmental defects in previous works. Those mutants, which were found as developmental determinants for cells other than touch receptor neurons, showed changes in touch receptor neuron development at a high frequency, which implied that many developmental and behavioral processes are regulated by a complicated combination of gene functions. The work also indicated that only a small portion of the markers, such as mec-7 and mec-4 genes, which are expressed in a touch receptor neuron-specific manner, and developmental mutants showed new biological insights into the causal relationships of development and behavior in C. elegans. The analysis revealed that additional genes are required to regulate the correct expression of touch receptor neuron-specific genes. The egl-44 (egg-laying defective) and egl-46 mutants had extra mec-7-positive cells in the head, which were transformed from FLP head sensory neurons. The lin-4 (abnormal cell lineage) mutants had ectopic mec-7-expressing cells in the mid-body, which were transformed from PVD mid-body sensory neurons. Loss-of-function mutations in lin-14 resulted in the loss of PVM neurons. ced-3 (cell death abnormality) and ced-4 mutants had extra mec-7expressing cells that were survivors from apoptotic cells in the PLM cell lineage. sem-4 (sex muscle abnormal) mutants had ectopic mec-7-expressing cells in the tail, presumably transformed from PHC neurons near PLMs. Although the additional mutants found in this work did not show apparent touch-insensitive phenotypes, they were needed for correct differentiation of touch receptor neurons. Thus, given that many gene markers and many mutations are used in combination, it is unsurprising that enormous amounts of information would be obtained. At that time, only classical gene expression markers were available. However, the introduction of the green fluorescent protein (GFP), which were originally obtained from a species of jellyfish (Aequorea victoria), as a neuronal visualization reporter for C. elegans mec-7 was enthusiastically appreciated by researchers in many fields of life science. 19), 20) Since the introduction of GFP as a biological reporter, many more fluorescent proteins with various fluorescent characteristics, including excitation/emission color spectra, have been developed, 21) allowing, for example, double labeling of cells and/or molecules of interest as the experiment requires. 22) Although the labeling of cells/molecules had been much improved by that time, the collection of mutants defective for genes of interest was lagging far behind because of the difficult and laborious experimental procedures that the process required. 23), 24) However, the introduction of a method by Mello et al. to generate transgenic strains of C. elegans expedited the analysis by fluorescent protein reporters and functional analysis by ectopic expression of genes of interest. 25) Because C. elegans bodies are transparent, they can be conveniently examined for fluorescence under the microscope, and thus they are useful for transgenic analyses. Transgenic strains for functional genomics Because transgenic strains have been readily available after year 1991, many researchers have worked with transgenic strains. We adapted the original methods developed by Mello et al. for our studies of functional analysis. A. Integration of transgenes into chromosomes. The original methods to isolate transgenic strains yield animals harboring extrachromosomal transgenes, usually called extrachromosomal arrays. Extrachromosomal arrays are often lost during development and between generations, and they cause mosaicism, which means loss of transgenes in subsets of somatic cells, and loss of transgenes in the progeny. The characteristics of extrachromosomal arrays themselves sometimes have an experimental advantage, e.g., for mosaic analyses, which utilize the mosaicism of transgenes as the random and differential presence or absence of proteins of interest, to determine whether phenotypes are caused by a cell-autonomous gene function. However, in many cases involving extrachromosomal transgenic strains, mosaic expression may result in, for example, partial rescue of the mutants and make experimental results unclear. To solve this problem, integration of extrachromosomal arrays into the chromosomes is useful. We used UV irradiation to integrate extrachromosomal transgenes into a chromosome. 26) After irradiation of transgenic strains with a UV crosslinker, which is used to fix DNAs and RNAs to nylon membranes for Southern and northern blotting, worms were allowed to produce F1 (daughter generation) animals. F1 animals were cultured in individual plates, and integrated strains were examined to determine whether the transgenic markers were segregated in the F2 (granddaughter) generation and later transmitted to all the progeny by inspection of the expression of fluorescent protein or other markers under a dissecting microscope. We were able to isolate strains that had stably integrated transgenes from up to 5% of parent extrachromosomal strains in this way. B. Analyses of the effects of ectopic expression of human disease-causal genes. After we obtained stable transgenic strains by integrating extrachromosomal transgenes, we applied the technique to a number of analyses. Two examples are shown here. We collaborated with Dr. T. Iwatsubo on the ectopic expression of ,-synuclein, a causal gene of familial Parkinson's disease and a major protein of Lewy bodies of sporadic Parkinson's disease. 27) ,-Synuclein is one of the best-known genes related to Parkinson's disease and is enriched in synapses and nuclei, as suggested by the name. There is no homologous gene for ,-synuclein in the C. elegans genome. We overexpressed ,-synuclein using promoters that are active in C. elegans dopaminergic neurons. 28) The ,-synuclein accumulated in the cell bodies and neurites of dopamine neurons. We did not find neuronal loss in these transgenic strains. It is known that dopamine is important for C. elegans behavior in response to the presence or absence of food. 29) ,-Synuclein transgenic animals showed behavioral abnormalities similar to those of dopamine neuron-deficient mutants and were rescued by the administration of dopamine in the culture. Thus, the overexpression of ,-synuclein in C. elegans dopamine neurons caused functional abnormalities within a shorter period than human disease progress. Additional transgenic strains were used to screen for modulators of ,-synuclein toxicity. 30) Transgenic strains expressing ,-synuclein in all neurons under the unc-51 promoter showed a weak Unc (uncoordinated) phenotype. We used 1,673 RNAi (RNA interference) clones to find the modifiers of the phenotype by looking for severe growth or motor defects. We found 10 genes contributing to the phenotype. Among these were four genes related to the endocytic pathway. The RNAi phenotypes were confirmed by crossing the transgenic animals with mutants. Furthermore, the accumulation of phosphorylated ,-synuclein was observed in neuronal cell bodies in an AP-2 (a clathrin adaptor protein) subunit knockdown, mimicking synucleinopathy (,synuclein-mediated neurodegenerative disorder). C. Screening for splicing factors. Another application of chromosomally integrated transgenic strains was to search for alternative splicing factors. It is important for multicellular organisms to have alternative splicing mechanisms to generate many more varieties of functional proteins than the number of genes in the genome by using distinct sets of exons from a gene. To address this issue, we collaborated with Dr. H. Kuroyanagi and Dr. M. Hagiwara to screen for C. elegans alternative splicing mutants. We chose the gene egl-15, which has mutually exclusive exons by alternative splicing and tissuespecific expression. 31) We constructed reporters indicating which exon was expressed by alternative splicing, enabling us to detect changes in the splicing pattern by color. We injected plasmids and obtained extrachromosomal transgenic animals harboring egl-15A fused to RFP (red fluorescent protein) and egl-15B fused to GFP. EGL-15B protein was mainly expressed in the hypodermis and neurons in the wildtype background, whereas EGL-15A protein was mainly expressed in muscle cells. We then integrated the transgene into a chromosome. The transgenic animals showed predominantly " red" fluorescence because of a larger volume of muscle cells. Then, we mutagenized the transgenic animals and searched for "green" worms by using a fluorescence-activated worm sorter. We obtained green mutant worms, and among them were mutants that had lesions in a gene, asd-1 (alternative splicing defective), a homolog of the human FOX family splicing factors. We found that asd-1 was partially redundant with fox-1 (a FOX ortholog) and alternative splicing patterns were gradually changed by the dosage of these genes. Isolation of gene knockout strains as a tool for reverse genetics Once mutant phenotypes have been described, researchers can perform transgenic rescue experiments to examine protein structure-function relationships. With the aid of transgenic techniques, the in vivo functions of wild-type and mutant sequences can be compared in C. elegans by examining the rescuing activity of the mutant phenotypes. Transgenes with a domain swapped with other genes, including human homologs, may also reveal the structure-function relationships of the molecules of interest. Thus, using mutants, we can infer more definitive causal relationships rather than mere correlation. C. elegans was the first animal model to have its whole genome sequenced. During the process, researchers were aware of the usefulness of the sequence information because sequencing revealed that many homologous genes, with which higher vertebrates share the same ancestral genes, are present in the C. elegans genome. Many laboratories tried to develop methods to isolate mutants that could be used for reverse genetics (see Fig. 1 legend). 23),32) However, in the early stages of those trials, many people recognized that it was very laborious to isolate mutants. To overcome this problem, we tried to develop efficient mutagenesis and screening methods, which eventually allowed us to obtain thousands of mutants for gene function analyses (so-called "tm mutants" named after the allele names of the mutants isolated in our laboratory). We preferred to use deletion mutants because deletions are easy to detect by PCR (polymerase chain reaction) and agarose gel electrophoresis and are more likely to disrupt gene functions than point mutations. A. ben-1 as both a humble and assertive gene. C. elegans O-tubulins are highly homologous with each other except in the C-terminal regions. 33),34) While working at Columbia University, I learned of the interesting phenotypes of ben-1 mutants published by the Chalfie laboratory; 33) wildtype animals are sensitive to an anti-tubulin drug benzimidazole and become sick, Dpy (dumpy) and Unc (uncoordinated movement), and eventually die. By contrast, ben-1 mutants are completely resistant to the chemical and paradoxically healthy on the media containing the drug. ben-1 mutants do not show any visible phenotype on their own, presumably because the ben-1 gene is compensated by the redundant function of other O-tubulin(s), and the mutants are completely healthy on the plates with or without the chemical (Fig. 2); the wild-type BEN-1 O-tubulin protein has dominant toxic effects on worm viability in the presence of the chemical. Moreover, the ben-1 gene is small in size, spanning only approximately 3 kb of the genome, and the only one gene that produces the Ben (benzimidazole-resistant) phenotype. We exploited the characteristics to isolate many mutations that showed resistance by random mutagenesis. 35) C. elegans researchers often mutagenize worms with EMS (ethylmethanesulfonate) to isolate new mutations. 10) However, EMS mutagenesis usually induces point mutations rather than deletions. Because genetic analyses usually require crossing between strains, it is more laborious to use mutants with a point mutation to examine genotypes. Thus, we chose a mutagenesis method with ultraviolet-activated trimethylpsoralen, which introduces crosslinks between neighboring nucleotides and results in deletions during DNA repair. 36) We mutagenized worms under various conditions, cul-tured worms on selection media, and isolated many alleles. Then, we mapped the genome lesions in and around the ben-1 gene by PCR and direct sequencing. In this way, we determined how to mutagenize the worms to isolate appropriate deletion mutations that were readily detectable with PCR and agarose gel electrophoresis. The ease of isolating ben-1 mutants also helped our research later (see below). B. Sensitive detection of deletion mutants by PCR. It was also necessary to detect deletion mutations within a large amount of wild-type DNA because random mutations occur at a low frequency in the gene of interest. PCR-amplified deletion mutant DNA fragments are shorter than those of wild-type animals. We developed a very sensitive PCR protocol called "stringent PCR". 35) We could preferentially amplify a small quantity of shorter (deletion mutant) targets over abundant longer (wild-type) targets, even if the shorter DNA is diluted 1:10,000 with wild-type DNA. When we detected mutant bands, we thawed frozen mutants and confirmed the deletion in growing progeny by sib Functional genomics of Caenorhabditis elegans No. 8] selection (Fig. 3). After isolating the mutants of interest in this manner, researchers can easily handle the mutants, because the genotype can be identified using only PCR and agarose gel electrophoresis even from a single animal, without sequencing every time. Simple knockout (KO) strains often serve as good biological materials for gene function analyses in addition to RNAi In many cases, scientists in the C. elegans research community have used deletion mutants as simple loss-of-function alleles. As had already been shown in mouse, 37) simple loss-of-function mutations in C. elegans often resulted in some phenotypes that were similar to those found by RNAi. A. Control of the engulfment of apoptotic cells by surrounding phagocytic cells. We conducted many interesting studies by generations and phenotyping of gene knockout strains. The signaling mechanisms underlying the signals from dying cells to phagocytes were examples of such analyses. We collaborated with Dr. D. Xue and Dr. X. Wang on the subject (Fig. 4). It is well known that apoptotic cells are removed by engulfment by neighboring phagocytic cells in C. elegans. 38) Extensive genetic analyses on this mechanism have been conducted. We isolated many mutants of genes that may be involved in this process. In C. elegans, as in other organisms including humans, phosphatidylserine (PS) is the key molecule of apoptotic cell removal. PS, which is located in the inner leaflet of the plasma membrane of healthy cells, is exposed to the cell surface of apoptotic cells, serving as the "eat-me" signal. The membrane receptor, ced-1 on the phagocyte membrane is responsible for finding dying cells and presumably detecting the eat-me signal. When the ced-1 gene is lost, many more cell corpses, which are dying but not yet removed, are visible than in wild-type animals because of the delayed removal of dead cells. 39) However, even in the ced-1 mutant background, cell corpses are removed eventually, although the process takes longer. A reasonable mechanism is that another pathway may transmit the presence of PS to phagocytes. Wang et al. showed that another membrane receptor, PSR-1, plays a role in detecting PS. 40) When the psr-1 gene is knocked out by a deletion mutation (tm469), the presence of cell corpses in worms is prolonged (55% longer than those of wild-type animals). The rescue activity of the transgenic psr-1 gene is dependent on the expression of phagocytes because expression of PSR-1 by the ced-1 promoter, which is expressed in phagocytes but not in the dying cells, is necessary. The psr-1 mutation in conjunction with other mutations of ced-1, ced-6 and ced-7 enhanced the corpse-prolonged phenotype, but in conjunction with mutations of ced-2, ced-5, ced-10 or ced-12 it did not. Since the PSR-1 protein interacts with CED-5 and CED-12, which are cytoplasmic signaling molecules, PSR-1 is thought to be a PS receptor for apoptotic cell engulfment. Mechanisms of externalization of PS to the outside of the membrane were explored. It is believed that localization of phospholipids is dependent on phospholipid scramblase. 41) Wang et al. showed that a mitochondrial factor WAH-1 promotes this externalization through a scramblase, as demonstrated by a deletion mutation of scrm-1 (SCRaMblase, tm698). 42) The mechanisms of asymmetric localization of PS on the dying cells were also explored. 43) We isolated deletion mutants for aminophospholipid translocases that may promote the inward movement of aminophospholipids such as PS. There are six homologs of the human phospholipid translocases (tat-1 to tat-6 (transbilayer amphipath transporters)). When worm germ cells treated with RNAi against tat-1 were labeled with annexin V, which binds to PS, faint staining was observed at the cell surface. RNAi against other tat genes did not produce the same phenotype. We used a tat-1 deletion mutant (tm1034) and a tat-3 deletion mutant (tm1275) to reproduce the experiment. Staining these mutants with annexin V resulted in a strong signal in the tat-1 mutant animals but never in the tat-3 mutant animals, confirming the RNAi experiments. Thus, the tat-1 gene underlies the restriction of PS to the inner leaflet of the plasma membrane in C. elegans. PS was able to induce removal of non-apoptotic cells. When six touch receptor neurons were labeled by GFP and examined, cells were more susceptible to loss in the tat-1 mutant background, suggesting that phagocytes sometimes attack cells that are not undergoing apoptosis if PS is on the membrane surface. B. Longevity regulated by the FoxO pathway. C. elegans is a popular organism for longevity analyses because of its short life cycle: 3 days to grow to adulthood and a total lifespan of 2 weeks. Direct screening for long-lived mutants is difficult because isolated mutant animals are too old to lay eggs and screening itself takes a long time. Furthermore, many unhealthy mutants show short-lifespans and it can be difficult to distinguish the different mechanisms affecting longevity. Thus, after finding the basic important pathways for longevity, it is better to screen for enhancers or suppressors of some preexisting longevity mutants with another phenotype, such as Daf (dauer formation-defective). Dr. P. Hu and Functional genomics of Caenorhabditis elegans No. 8] colleagues tried to isolate such genes systematically; for example, eak (enhancers of akt-1), which may be among the daf-16 (a homolog of FoxO) pathway genes and is involved in longevity in parallel to akt-1 (akt kinase family). After identifying an interesting gene, eak-7, it is convenient and reliable to have another allele of the eak-7 mutant to test whether the phenotypes obtained by the forward genetics are reproducible by null mutations. 44) An eak-7 deletion mutant (tm3188) enhanced the dauer arrest phenotype. This phenotype was suppressed by the daf-16 mutation, suggesting that the EAK-7 protein inhibits the DAF-16 protein. eak-7 mutants live longer than wild-type animals, as expected from the daf-16 inhibition. C. Mechanistic insights into RNA interference. RNAi became a very popular experimental technique among the C. elegans research community after an exciting paper from the Fire and Mello laboratories was published in 1998. 45) Today, researchers working in many different research fields use the RNAi method to assess the effects of downregulation. We collaborated with Dr. C. Mello to explore the mechanisms of RNAi using our mutants, because RNAi knockdown of genes needed for RNAi is often difficult to interpret. We isolated a number of mutants of genes for DCR-1 (Dicer related) interactors. 46) Among the interactors, PIR-1 (phosphatase interacting with RNA/RNP) and DRH-3 (Dicer related helicase) were essential proteins for RNAi. These proteins were shown to play roles in processing double-stranded RNAs. We also obtained deletion mutants for all 27 Argonaute (AGO) genes and examined the roles of each clade in RNAi and related gene-silencing mechanisms. Single-and multiple-AGO mutants showed several distinct phenotypes in relation to the corresponding genetic pathways: chromosome segregation, fertility, and RNAi. 47) Several AGO proteins interacted with amplified siRNAs and microRNAs, suggesting that these AGO proteins are limiting factors of both microRNA and exogenous and endogenous RNAi pathways. D. Some examples concerning vesicular trafficking machinery. We have been working on vesicular trafficking. The GON-1 protein is an ortholog of human ADAMTS9 and is necessary for C. elegans gonadal cells to proliferate and migrate. The gon-1 gene encodes a secretory matrix metalloprotease that digests the extracellular matrix and helps gonadal migration. 48) We found that gon-1 (tm3146) mutants show not only a sterile phenotype but also the accumulation of abnormal membrane structures in the cytoplasm. We found that gon-1 is necessary for proper vesicular trafficking of many proteins, including secretory proteins, membrane proteins, and GPI-anchored proteins. 49) We examined whether ADAMTS9 has a similar role in human cells by knocking down the gene in HEK293 cells. This was the case, and vesicular transport from the endoplasmic reticulum to the Golgi apparatus was severely impaired. Interestingly, the GON domain located in the C-terminus, but not the metalloprotease domain, plays a major role in vesicular trafficking. We also found that transfection of the C. elegans GON domain cDNA into HEK293 cells rescued the trafficking defects by knockdown of the ADAMTS9 gene. Thus, the GON domain is important for the secretory pathway in both C. elegans and humans. E. Stem cell maintenance mechanisms. To study disease mechanisms, we chose a gene bcl-7 (mammalian BCL (B cell lymphoma) gene homolog), which is homologous to human BCL7 gene family members (BCL7A, BCL7B and BCL7C). BCL7B is one of 28 genes that are deleted in Williams-Beuren syndrome, but little is known about its role, although clinical reports describe a high incidence of blood malignancies. 50) We isolated a deletion mutant bcl-7 (tm5268) and analyzed the phenotypes. We found that bcl-7 is necessary for the maintenance of the stem cell characteristics of seam cells, which are stem cell-like cells in the C. elegans postembryonic lineage. 51) Furthermore, the bcl-7 mutants are sterile. Interestingly, bcl-7 mutants have enlarged nuclei. We used a human cancer cell line, Kato III, and found that knocking down of human BCL7B results in a similar enlargement phenotype with the appearance of malignant morphological changes to the nuclei. Because C. elegans has only one bcl-7, it was a convenient model in which we explore the basic function of the BCL-7 protein to gain insights into the human BCL7 family. F. Some discrepancies between phenotypes found in deletion mutants and RNAi. We sometimes encounter discrepancies between phenotypes observed by RNAi and gene knockout strains. Theoretically, it is conceivable that some RNAi phenotypes are different from those generated by gene knockouts. For example, the redundant function of two homologous genes may indicate that removing only one gene in the genome does not result in any phenotype, but double mutants show some phenotypes. However, RNAi downregulates transcripts from both genes, resulting in visible phenotypes. In different contexts, some genes show very weak or no phenotypes by RNAi but strong phenotypes by gene knockout; such phenomena are described for many genes expressed in nerve cells, especially when RNAi is administered by feeding. To solve this problem, researchers sometimes use transgenic RNAi with neural promoters, which allows dsRNA to be expressed in neurons and to act cell-autonomously, 52) or overexpress the sid-1 (systemic RNA interference defective) gene with a neuronal promoter, 53) which helps the uptake of dsRNA by neuronal cells. In these situations, gene knockdown animals often show apparent phenotypes. It is also expected that maternal-effect genes may show distinct phenotypes between RNAi and gene KO. Homozygous mutants from heterozygous parents (by self-progeny) show weak phenotypes during the early embryonic stages in the gene KO strains and, thus, can be maintained as heterozygotes, whereas sometimes they show severe lethal phenotypes by RNAi. Because feeding parental animals with RNAi clones causes the degradation of maternal mRNAs in the embryo, this is the expected difference between RNAi and gene KO strains. Another difference between RNAi and the mutants is that the phenotypes observed by RNAi are sometimes variable among experiments, from weak phenotypes to severe lethality (many examples are described in WormBase), depending on the researchers (differences of delivery methods of dsRNA) and the RNAi clones (the length and regions of inserts). Other mechanisms may be envisaged. Gene KO may not represent all the possible phenotypes in which a gene is involved; when an essential gene is involved in two or more developmental stages, phenotypes that should appear in later stages are more difficult to observe. In other cases, some loss-offunction mutations may be compensated by homologous genes. The possibilities that certain phenomena may use more than two pathways should also be considered (typically, daf mutations; see below). To examine such genetic pathways, more sophisticated and intensive approaches are needed, reminiscent of the complicated situations of multifactorial diseases in humans. We ourselves experienced some discrepancies in phenotypes between RNAi and gene knockout strains. For example, the vps-45 gene did not have a visible phenotype with RNAi clones but showed a temperature-sensitive lethal phenotype with gene knockout strains. 54) Also, vps-33.1 and vps-33.2 knockouts showed phenotypes different from those previously reported by RNAi clones. 55),56) The reasons why KO or RNAi of some genes induces different phenotypes are often difficult to explain. 7. Methods to help analyses of gene functions in C. elegans As exemplified above, sometimes we should use multiple and more sophisticated techniques to reveal gene functions. We have tried several useful techniques, as shown below. A. Single/low-copy integration of extrachromosomal transgenes. The gene expression of C. elegans transgenic strains is silenced in germ cells. This occurs because multicopy transgenes are recognized as repetitive sequences and inactivated. Transgenic animals generated by microinjection of DNA solution into the germline syncytium usually contains multicopy repetitive extrachromosomal DNAs that are silenced in the germline but expressed in the somatic cells. 57) As mentioned above, multicopy transgenes are silenced in the germline. To solve this problem, a single-copy transgene can be integrated into the chromosome via Mos1 transposon excision. 58) We have developed an easy method for integrating singleor low-copy transgenes. 59) We used a ben-1 and vps-45 double mutant strain, which shows a benzimidazole-resistance and temperature-sensitive lethality phenotype as mentioned above. We added rescue DNA plasmids with the ben-1 and vps-45 genes when we generated transgenic strains. The extrachromosomal multicopy transgenic animals became wild-type for both the ben-1 and vps-45 mutations, and were sensitive to benzimidazole but viable at 20°C. If these transgenic strains are UV-irradiated and cultured on selection media, a small fraction of animals have single/low-copy transgenes in some chromosomes. Many of the transgenic animals retain the wild-type ben-1 gene, showing sensitivity to benzimidazole. In addition, only animals that maintain the vps-45 rescue transgene can survive. This can occur only when short transgenic DNAs that are devoid of the ben-1 fragment but retain the vps-45 fragment are integrated into a chromosome. Thus, by culturing transgenic animals on selection media at 20°C, researchers can easily obtain single/low-copy transgenic strains by simply waiting for animals to grow on the media. Although this selection yields random integration again, we have now successfully integrated single/low-copy transgenes into defined loci using the genome editing method. 60) We used the CRISPR/ Cas9 method to cut a chromosomal integration site and a transgene sequence. In this way, we could integrate single/low-copy transgenes into the loci of interest, making it easier to generate multiple variations of mutations and transgenes to cross with other mutants or transgenic strains. Single-copy integration is also useful to avoid overexpression of toxic transgene products. 61) B. A conditional KO system. In C. elegans, as mentioned above, lethal mutants are often difficult to use for behavior analyses. However, conditional knockout strains are routinely used in mice. 62) We suspect that the use of conditional knockout strains is unpopular among C. elegans researchers because most C. elegans transgenic strains are extrachromosomal and multicopy. Multicopy transgenes are not suitable for conditional knockout; even if transgenes are floxed (loxP sequence from bacteriophage P1 is inserted and can be recombined between two loxP sites by Cre recombinase from bacteriophage P1), only some of the copies are removed by recombinases, resulting in the failure of knockout. After succeeding in single-copy integration, we next tried to perform conditional knockout experiments. 61) Single-copy integrated floxed DNA fragments could be excised cleanly by crossing with Cre recombinase-expressingtransgenic strains driven by spatiotemporally specific promoters, enabling a conditional knockout. Although it was easy to remove the transgene in a spatiotemporally specific manner, we found that conditional experiments in C. elegans must be performed carefully. C. elegans grows very fast: it takes only 3 days for a fertilized egg to develop into an adult animal, and therefore, some proteins may not be removed in some cell types and tissues, while the same proteins are removed almost completely in other cell types. Accordingly, some phenotypes may appear incomplete. Applications should be carefully chosen with this limitation in mind. C. A CALI system for stable proteins. Instead of removing the genome sequence, we found that some proteins may be better analyzed by chromophore-assisted light inactivation (CALI). In C. elegans, mutants of many genes can be rescued by the wild-type transgene. We often find that GFPfusion genes can also rescue the phenotypes, albeit partially. In such cases, we can replace GFP with KillerRed and rescue mutant phenotypes. KillerRed, a fluorescent protein-related protein, absorbs green light and generates superoxide: when overexpressed it can kill the expressing cells. 63) In our hands, it was difficult to kill cells with KillerRed. However, inactivating fusion protein was easy and reproducible. Because the absorbance maximum is in the wavelength range of green light and the excitation light itself rarely affects cell function, protein function can be safely and easily switched on and off. For example, we expressed a KillerRed fusion protein in mutant animals with severe phenotypes; gon-1 mutants and bcl-7 mutants. 49),51) We rescued the gon-1 mutant with transgenic expression of a KillerRed-GON protein. Using this strain, we investigate how the secretion process is regulated by the GON-1 protein. Another gene, bcl-7, is a homolog of human BCL7, and bcl-7 deletion mutants show sterile phenotypes because of abnormal somatic and germ cells. We rescued bcl-7 mutants completely with transgenes of a KillerRed-BCL-7 fusion and could inactivated the BCL-7 by illuminating the strain with a green light. Thus, we were able to examine when the protein is required for proper gonadal cell development. Because proteins appeared completely inactivated using this method, the slow turnover rate of the preexisting proteins of targeted genes is no longer a limitation. D. Creation of new balancers that enable more reliable description of mutant phenotypes. We isolated many deletion mutants and found a certain portion of mutants showing lethal or sterile phenotypes (NBRP for the nematode: http://shigen. nig.ac.jp/c.elegans/index.jsp). Because C. elegans animals grow and proliferate quickly as hermaphrodites, such mutants need to be balanced: otherwise, mutant animals are diluted by the segregated wildtype animals, which grow faster than mutants. To maintain lethal or sterile mutations, researchers often use balancer chromosomes, which are associated with some phenotypes and are used as heterozygote with mutant chromosome so that healthy heterozygous animals grow as a large population. 64) Unfortunately, most C. elegans balancer chromosomes are translocations generated by .-ray irradiation, which give rise to aneuploidy. Theoretically about half of the self-progeny die from the balancer itself. This is very inconvenient for analyzing, for example, embryonic lethal strains because the developmental stages at which the mutant phenotypes of interest and the developmental arrest by the balancer aneuploidy occur are very close together. There are a few crossover balancers created by .-ray irradiation, which have complex rearrangements within the same chromosome and, thus, do not show any aneu-ploidy. 64) We used the CRISPR/Cas9 technique to create crossover balancers. 65) We induced chromosomal inversions twice in two overlapped regions in the same chromosome, which are suitable both for stable maintenance of covered mutant alleles and for phenotype analyses. We have now generated such crossover balancers to cover the whole C. elegans genome and labeled the balancers with fluorescent protein reporters (Dejima et al., manuscript in preparation). We expect that with these balancers, mutations with lethal phenotypes will be analyzed more readily and exactly. Discussion A. Homology analyses are useful, but we sometimes need more. Genome sequence information is essential for medical and life science studies. We routinely search for homologies among genes and their products, and the findings are very useful to understand gene functions. In light of our limited experiences in functional genomics research, sometimes we need to know more. To reveal the disease susceptibility genes, an enormous number of studies are being performed, with sophisticated algorithms. One reason that makes it very difficult to examine the human genome is the high variation of genetic backgrounds among individuals. Even if a gene is related to some diseases, odds scores by statistical analyses are often low. Model organisms shed light on such enigmas. We found that an amino acid change in Argonaute proteins removes their catalytic function and changes the role of the subfamily into RNA receptors from RNase. 47) DCR-1 was found to digest DNAs instead of dsRNA after it was cut by a caspase, CED-3. 66) There are many cases in which alternative splicing changes not only gene expression but also protein function: for example, the ubiquitously expressed protein DAF-16 (a homolog of mammalian FOXO) produces different phenotypes when different alternatively spliced exons are removed. 67) The function of GON-1, a homolog of human ADAMTS9 metalloprotease, made us recognize that an important domain (GON domain) in the C-terminus, which is not related to protease but plays a critical role for cells to normally secrete proteins, has been ignored because the metalloprotease domain is so obvious. 49) An rSNP (regulatory single nucleotide polymorphism, which has a variation in a gene regulatory sequence but not a protein coding sequence) of the human gene ADAMTS9 is regarded as a susceptibility gene for type 2 diabetes, but a knockout mouse strain was reported as fetal lethal. 68) gon-1 null mutants show a number of severe phenotypes, including sterility, but we also found that certain genes, if downregulated, suppress even such severe phenotypes (Yoshina et al., in preparation). Medical research usually targets diseases with some symptoms detected after birth. In an extreme idea, some of the strains we are using may have such mutations in the vicinity of the same linked chromosome from the start and may not simply be " wild-type" strains. In the same way, humans have many combinations of variations, and this makes problems very difficult to solve. Interestingly, a small fraction of adult population has severe and high-penetrant Mendelian childhood disease mutations with no clinical manifestation, suggesting that they may harbor protective genetic variants. 69) B. Integrated genetics and future direction. We have been using a model organism to understand gene functions (Fig. 1). What fundamental insights can we expect to obtain in the future to solve questions in the medical and life sciences? For example, it is intriguing to consider whether it is possible to simulate all the behavior of C. elegans as an initial move toward understanding the human mind as a connectome, perhaps as originally proposed by Brenner. 9), 70) We have the wiring diagram of C. elegans, showing how synaptic connections are formed among neurons. 8) The C. elegans research community is analyzing the C. elegans body to reveal everything about the worm and to try to use the information to understand more general mechanisms of life, including development, behavior, cell biology, and even human diseases. However, we sometimes find unexpected phenomena beyond the inference from known information. The reasons why biological analyses are so complicated exist in the organisms themselves. Biological phenomena are based on multilayered structures: molecules, cells, tissues and whole bodies in various time courses, which make biological analyses complicated. In addition, a similar phenomenon may be caused by signal transduction between distinct molecules, distinct tissues, and even by distinct environments. For example, with classical mutant analyses, we know that the phenomenon of dauer formation consists of at least four distinct pathways. 71) This shows us that a phenomenon, or seemingly undistinguishable multiple phenomena, may be caused by completely independent pathways and should be examined carefully. We also found that a Wnt molecule may cause apparent opposite effects for bcl-7 mutant phenotypes, which showed a haploinsufficient phenotype. 51) This is reminiscent of the fact that Williams-Beuren syndrome is caused by haploinsufficiency of chromosome 7q11.23. 72) It is a common idea that gene expression is regulated in a temporally-specific and tissue/cell-type-specific manner, so a single line of evidence may not give a complete picture of the physiological and pathological states of the body. We have been using transgenic expression, RNAi, and gene knockout strains in addition to mutants isolated by forward genetics. Moreover, we have shown that it is now possible to obtain drastically gene-modified strains, such as artificial structural changes of the chromosomes. 65) To elucidate all the gene functions in a model organism will require integrated genetics, which is a combination of forward and reverse genetics, a systematic and large set of reduction-of-function studies using RNAi clones 73) or gene knockout strains, which are or will be available for C. elegans in the near future as they are in yeast. 74) Moreover, we should combine gainof-function transgenic studies with reduction-offunction studies. We also need to combine enhancer screening and suppressor screening to reveal possible gene networks or genetic pathways. Recently, forward genetics with next-generation sequencing has enabled researchers to identify C. elegans mutations quickly and easily. 75) To understand the genetic pathways involved in all the layers of the different biological processes in C. elegans, we need to examine genotype-phenotype relationships for every possible phenomenon, for every possible developmental stage and every possible environment. In addition, although the overall structures of C. elegans and humans are very different, the genetic pathways are partly similar and information relating to those genetic pathways is valuable for understanding the nature of health and diseases in humans.	2018-04-03T05:54:28.462Z	2017-10-11T00:00:00.000	{ "year": 2017, "sha1": "ec0bdfaf2218d7780f7485f0db251e0842a86288", "oa_license": null, "oa_url": "https://www.jstage.jst.go.jp/article/pjab/93/8/93_PJA9308B-03/_pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "ec0bdfaf2218d7780f7485f0db251e0842a86288", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
119087144	pes2o/s2orc	v3-fos-license	Theory of Photoinduced Phase Transitions Theories of photoinduced phase transitions have developed along with the progress in experimental studies, especially concerning their nonlinear characters and transition dynamics. At an early stage, paths from photoinduced local structural distortions to global ones are explained in classical statistical models. Their dynamics are governed by transition probabilities and inevitably stochastic, but they were sufficient to describe coarse-grained time evolutions. Recently, however, a variety of dynamics including ultrafast ones are observed in different electronic states. They are explained in relevant electronic models. In particular, a coherent lattice oscillation and coherent motion of a macroscopic domain boundary need appropriate interactions among electrons and lattice displacements. Furthermore, some transitions proceed almost in one direction, which can be explained by considering relevant electronic processes. We describe the history of theories of photoinduced phase transitions and discuss a future perspective. Introduction Nonequilibrium phases can be generated from an equilibrium one by external stimulations. Photoirradiation is one of them and sometimes allows the emergence of a new phase, which cannot be reached by simply changing temperature or pressure, because the energy of a photon is much higher than thermal energies. Although these phases finally relax to the starting equilibrium phase, they are locally stable, separated from the initial one by a substantial free-energy barrier (except for the transition from a Mott insulator to a metal, as will be mentioned later). When the relaxation time is long enough, the properties of this nonequilibrium phase can be studied by laser spectroscopy techniques, which have recently made great progress. Thus, photoinduced phase transitions have been studied extensively, both experimentally and theoretically. [1][2][3] Here, we review progress in theories for such phase transitions. Photoirradiation creates electrons and holes, or excitons if they are bound. In most of the cases, they are accompanied by local structural deformation. To develop it spatially, some cooperativity is needed. One would ask first what kind of interaction possesses such cooperativity. Section 2 introduces classical models, in which photoexcited states proliferate in some conditions. Ground and excited states are simply assigned to Ising variables, so that the electronic transitions are statistically described on the basis of diabatic potentials. Section 3 first introduces adiabatic pictures for electronic models, which are useful in discussing necessary conditions for the spontaneous growth of photoexcited states. Transition dynamics do not always proceed adiabatically. Next, this section introduces the electronic dynamics de- * E-mail address: kxy@ims.ac.jp rived from the time-dependent Schrödinger equation and shows their relevance to experimental findings. Section 4 concludes and presents future issues. Linear chain system In order to clarify how the local structural distortions lead to global ones, a theory based on stochastic processes was proposed at first. Hanamura and Nagaosa introduced a simple model: 4) and \| g l g l \| + \| e l e l \|= 1 , where E FC is the Franck-Condon excitation energy, Q l the relevant displacement that dominantly interacts with an electron in the photoexcited \| e l state at the lth molecule and is treated as a classical variable, and √ SQ l the Stokes shift in this state. The coupling strength between the displacements of the lth and l ′ th molecules is denoted by K ll ′ with its diagonal element K ll set at zero. Linear chain systems are first investigated in ref. 4. Suppose that electrons on m consecutive molecules (e.g., on sites 1 to m) are in the excited state, and others are in the ground state. Each molecule is located on the respective equilibrium position. This state is governed by the Hamiltonian H (m) g . Then, consider that an electron on a neighboring molecule (e.g., on site 0) is excited. The corresponding molecule is now away from the equilib- . The difference between these Hamiltonians is given by Then, an interaction mode q 0 (on site 0) is introduced, which is a linear combination of the displacements in the state with m excited molecules. The diabatic potentials for H are drawn as a function of q 0 (Fig. 1). They are displaced parabolas. The relative stability is determined by the minima of the two parabolas where q g and q e give the respective minimum points, and by is satisfied, the excited (ground) state at the 0th molecule is unstable and would decay into the ground ("excited") state. Otherwise, the smaller one of E g and E e corresponds to the stable state, and the larger one to the metastable state. The analysis based on these relations reveals how a cluster of photoexcited molecules is stabilized by the electron-lattice interaction, whose dimensionless strength is the lattice relaxation energy S divided by E FC , and by the coupling K among the displacements. A similar analysis also explains how a molecular excitation far from the cluster is attracted to the cluster. Mean-field theory Nagaosa and Ogawa have extended this theory to general dimensions and considered the optical and thermal transitions. 5) When the relaxation of the interaction mode Q l is rapid, the system is always in equilibrium with a given electronic state. These modes can be integrated out to reduce the original model (1) to the Ising model, where the spin-up and spin-down states correspond to the excited and ground states: where the field h, the exchange J ll ′ , and the phonon dispersion ω k are given by with R l being the lattice vector of the lth molecule, and ω 2 k=0 = 1 − K. The intersite couplings are treated in the mean-field approximation. For instance, the thermal transition probabilities were assumed to be the product of the attempt frequency and the activation factor, P th , where E c is the energy of the intersection of the two parabolas. The stochastic dynamics of the system under and after the optical pumping are investigated on the basis of the rate equation for the density of the excited states n e (t), which is a function of time t. When stable and metastable states are present, it is written as d dt n e (t) = I(t) + P th g→e [1 − n e (t)] − P th e→g n e (t) , (8) where I(t) is the optical pumping term, being positive to create the e state or negative to create the g state. The activation energies in the exponents of the thermal transition probabilities are quadratic functions of n e (t) in the mean-field approximation. The time evolution of n e (t) after n e (0)=0 depends on the strength I 0 and duration t 0 of the optical pumping. When I 0 is less than a threshold value I th , however long the optical pumping is applied, n e (t) does not exceed a certain value. After the pumping is switched off, the system always relaxes to the initial state n e (0)=0. When I 0 is greater than I th , on the other hand, n e (t) monotonically increases without saturation for 0 < t < t 0 . After the pumping is switched off, n e (t) relaxes to the final value n f =0 for t 0 < t cr and to n f =1 for t 0 > t cr . The critical value of the duration time t cr , above which the system is switched from n i =0 to n f =1, is inversely proportional to I 0 for large I 0 and diverges as t cr ∼ (I 0 − I th ) −1/2 near the diverging point I th . The relaxation times are also discussed. The theory is applied to photoinduced structural transformations in polydiacetylenes. Other theories The theories above are applied to systems with weakly interacting components to which thermal fluctuations are so dominant that the transition probability is a key quantity. Spin-crossover complexes are such examples, where spins indirectly interact with each other through couplings with lattice displacements. The Hanamura-Nagaosa theory is extended to systems with internal degrees of freedom, e.g., spin-crossover complexes that are composed of dimers and show two-step transitions. 6) In general, stochastic dynamics can be treated in classical models with master equations [7][8][9][10] and Monte Carlo simulations. 7, 10-12) Domino effect Photoinduced dynamics are sometimes regarded as domino effects. This description is inappropriate for systems showing coherent dynamics, but gives a hint for the mechanisms of photoinduced phase transitions. Koshino and Ogawa theoretically show a domino-like structural transformation in a one-dimensional model composed of localized electrons and classical lattice displacements: 13) where u j and ∂u j /∂t denote the jth displacement and velocity, respectively, k ij the elastic constants, ǫ the energy difference between the two electronic states at u j =0, t 0 the overlap integral between them, and γ the electronlattice coupling constant. After a photon is absorbed by an electron at a site, the lattice locally relaxes under friction to the minimum on the adiabatic potential. Then, after a photon is spontaneously emitted, three evolution patterns are shown to exist. When the elastic couplings k ij are too weak, the local structural distortion remains locally. When they are too strong, the initial local distortion is pulled back by surrounding lattice displacements to the position before the absorption. Only when they are intermediately strong and short-ranged will the initial local distortion trigger a global structural transformation as the domino effect (Fig. 2). Adiabatic relaxation path Photoinduced phase transitions are often related to multistability, where different electronic phases are stable and metastable. Such stability may be manifested by the presence of a first-order phase transition induced by changing temperature or pressure. The metastable phase may be hidden in equilibrium so that it is manifested only by photoirradiation, which gives a much higher energy to the system than thermal energies. The photoinduced phase transition dynamics depend on the electronic state and especially on how the initial state is prepared. For this purpose, relevant itinerant-electron models are necessary. They have finite transfer integrals between neighboring molecular orbitals. They are off-diagonal elements of the model Hamiltonian, giving transition amplitudes. In contrast, the stochastic dynamics within classical statistical models are governed by transition probabilities, which are often given by Boltzmann factors satisfying detailed balance. Organic molecular tetrathiafulvalene-chloranil (TTF-CA) crystals are the most intensively studied. The donor TTF and acceptor CA molecules are alternately stacked along the most conducting axis. At low temperature or under high pressure with contraction, these molecules become ionic due to the long-range Coulomb interaction. 14) Otherwise, they are neutral due to the large difference between the ionization potential of the donor molecule and the electron affinity of the acceptor molecule. In the ionic phase at ambient pressure, these molecules are dimerized. Both ionic-to-neutral and neutral-to-ionic transitions are photoinduced, as shown by the optical reflectivity. 15) It is found that photons with 2.3 eV corresponding to intramolecular excitations at TTF sites are much more efficient for generation of neutral domains in the ionic background than those with 0.6 eV corresponding to intrachain charge-transfer excitations. 16) The latter clearly have a threshold in the photoexcitation density, below which a macroscopic neutral domain cannot be generated. Thus, the photoinduced dynamics are very sensitive to the initial condition of the electronic state, from which the lattice relaxation starts. These observations have motivated theoretical studies based on itinerant-electron models. Huai et al. have clarified the adiabatic relaxation path from the Franck-Condon-type photoexcited state above the ionic ground state to the formation of a large neutral domain, using a one-dimensional extended Peierls-Hubbard model with alternating potentials (and the interchain interaction mentioned later): 17) where c † l,σ (c l,σ ) is the creation (annihilation) operator of an electron with spin σ at site l, n l,σ = c † l,σ c l,σ , n l = n l,↑ + n l,↓ , δn l = n l − 2 for odd l, δn l = n l for even l, q l is the dimensionless displacement of the lth molecule along the chain from its equidistant position. The distance between the lth and (l + 1)th molecules is given by d l,l+1 = d 0 (1 + q l+1 − q l ), where d 0 is the average intermolecular distance. The donor and acceptor molecules are located at odd and even sites, respectively. The nearest-neighbor repulsion strength is assumed to depend nonlinearly on the intermolecular distance as where V 0 is for the regular lattice, and β 1 (β 2 ) is the linear (quadratic) coefficient. The parameter t 0 denotes the transfer integral, ∆ the level difference between the neighboring orbitals in the neutral limit, and U the onsite repulsion strength. The elastic energy is expanded up to the fourth order: the parameters S 1 and S 2 are the linear and nonlinear elastic constants. In the neutral phase, the orbital of the donor molecule is almost doubly occupied, while that of the acceptor molecule is almost empty [ Fig. 3(a)]. The total charge of the donor molecule at site l is +(2 − n l ) = −δn l , while that of the acceptor molecule is −n l = −δn l . In the ionic phase, both orbitals are almost singly occupied [ Fig. 3(b)]. The values of the eight parameters are so determined that they reproduce the ab initio estimation of the transfer integral, the ionicity in the ionic phase, that in the neutral phase, the degree of dimerization in the ionic phase, the energies and the relative strength of the charge-transfer absorption peaks in the ionic phase, and the charge-transfer absorption energy in the neutral phase. The lattice relaxation path starting from the Franck-Condon state with a single charge-transfer exciton is assumed to be described by a domain, where (−1) l q 0 denotes the uniform dimerization in the ionic state, ∆q the amplitude of a local distortion induced by an excited domain, θ the width of the domain boundary, and l 0 the domain size (Fig. 4). An interchain interaction between displacements is introduced to confine the domain in the ionic background: where q l denotes the displacement of a central chain surrounded by ionic environmental chains, and K i the 2ith coefficient with respect to the distortion in the central chain. The adiabatic energy surface of the ground state and that of the first excited state are drawn as a function of the amplitude ∆q and the size l 0 of the domain with the width of the domain boundary θ optimized. 17) The energy surface of the ground state has a plateau corresponding to a neutral domain separated from the ground state by a low potential barrier. In the energy surface of the first excited state, a local minimum corresponding to the charge-transfer exciton appears around the Franck-Condon state. Another local minimum appears and corresponds to a neutral domain, which are separated from the Franck-Condon state by such a high potential barrier that the absorption of a single photon with 0.6 eV cannot overcome it. The domain is stable only when its size exceeds a critical value because the energy required for creation of the two domain boundaries makes a small domain unstable in the excited state. The spin-and chargedensity distributions of the ground and excited states with a neutral domain are very similar: they are different only at the domain boundaries. Because a single chargetransfer exciton cannot trigger the formation of a macroscopic neutral domain, a large excess energy is needed at the very beginning of the relaxation to overcome the high barrier, for charge-transfer excitons to proliferate, and finally to form a macroscopic neutral domain. The process described above is only for the early nucleation stage of the photoinduced macroscopic transformation. Transition dynamics in TTF-CA Along with the progress in experimental techniques, the pump-probe reflection spectroscopy acquires higher time resolution. Ultrafast optical switching from the ionic to the neutral state is observed. 18) Neutral strings are produced within 2 ps after a resonant excitation of the charge transfer band at 0.65 eV polarized along the stacking axis. Just below the transition temperature, a macroscopic conversion from the ionic to the neutral states is achieved within 20 ps. It is accompanied with coherent motion of the macroscopic neutral-ionic domain boundary with a much longer period than that of the optical lattice oscillation. Theoretically, real-time dynamics need to be calculated for further understanding. First, results are obtained by solving the time-dependent Schrödinger equation of an electron-lattice model without electronelectron interaction, where two charge-density-wave states with opposite polarizations are degenerate. 19) The lattice part is treated classically. The photoconversion is shown to be nonlinear. During the photoconversion process, macroscopic oscillations are observed, which are essentially different from the linear modes in equilibrium. Then, by adding the kinetic energy of the displacements, with the lth molecular mass m l , to the model (10), Miyashita et al. have calculated real-time dynamics of ionicity and dimerization within the unrestricted Hartree-Fock approximation after the occupancy of orbitals in the ionic ground state is initially changed. 20) As the number of orbitals whose occupancy is changed increases, i.e., with increasing photoexcitation density, more neutral domains are created. Above the threshold excitation density, the neutral phase is finally achieved. After the photoexcitation, ionic domains with wrong polarization generally appear, which reduce the correlation length of the staggered lattice displacement. As the degree of initial lattice disorder increases, more solitons appear between the ionic domains with opposite polarizations, which obstruct the growth of neutral domains and slow down the transition. From the Fourier transform of the ionicity as a function of time, three characteristic time scales are observed: rapid charge-transfer excitations corresponding to the peak energy of a chargetransfer exciton in the random-phase-approximation spectra, slow charge-transfer excitations strongly coupled with optical lattice oscillations, and much slower excitations due to the collective motion of neutral-ionic domain boundaries. The last corresponds to the experimentally observed, coherent motion of the macroscopic neutralionic domain boundary, 18) and is shown to be sensitively affected by collisions with the solitons. When the fraction of conversion from the ionic to the neutral phases is plotted as a function of the number of excited electrons, the curve evolves with time (Fig. 5). Immediately after the photoexcitation, it is close to a linear function. Above the threshold number, the fraction monotonically increases with time until the photoinduced phase transition is completed. Below the threshold, the fraction decreases and eventually becomes very small. This result is consistent with the experimental finding. 18) More recently, experimental results have been summarized covering both the ionic-to-neutral and the neutralto-ionic transitions with different excitation energies, excitation densities and temperatures. 21) The ionic-toneutral transition induced by the resonant excitation of the charge-transfer band proceeds with (1) initial formation of a confined one-dimensional neutral domain, (2) multiplication of such domains to semi-macroscopic neutral states by 20 ps, and (3) evolution in the direction normal to the sample surface. The dynamics of the neutral-to-ionic transition that is induced by a similar excitation are clearly different from the above. Although one-dimensional ionic domains are initially produced by lights, they decay within 20 ps even if the excitation density is high. The initial conversion fraction is a linear function of the excitation density. Thus, we are motivated to incorporate a pulse of oscillating classical electric field, with amplitude E ext and frequency ω ext for 0 < t < 2πN ext /ω ext with integer N ext , into the Peierls phase of the transfer integral. The time-dependent Schrödinger equation is solved again for the one-dimensional extended Peierls-Hubbard model with alternating potentials. Now, the frequency, the amplitude and the duration of the pulse can be explicitly and independently varied. The dynamics of the ionic-to-neutral transition are indeed qualitatively different from those of the neutral-to-ionic transition. Difference between I-to-N and N-to-I transitions When the dimerized ionic phase is photoexcited, the threshold behavior is observed again by plotting the final ionicity as a function of the increment of the total energy, i.e., as a function of the number of absorbed photons (Fig. 6). 22) Above the threshold photoexcitation density, the system finally enters a neutral state with equidistant molecules. This is the case even after the electric field is switched off. Once produced by the field, a neutral domain grows spontaneously. Its growth cannot be stopped by switching off the field. When compared with the threshold behavior derived from the rate equation, 5) we do not observe a quantity corresponding to the threshold intensity I th . Even if the amplitude of the pulse is small (i.e., even if the pulse is weak), a transition seems finally achieved by setting the pulse duration long enough. The deterministic dynamics are more efficient than the stochastic ones because of the restricted energy dissipation. When compared with the domino effect, the motion of the neutral domain is similar in that its growth is spontaneous once triggered. However, it is quite different from the motion of dominoes under strong friction on the adiabatic potential, 13) which is far from being coherent. For a given amount of the increment of the total energy, the lattice dynamics relative to the charge dynamics in the ionic-to-neutral transition depend on the character of the pulse. When the pulse is strong and short, the charge transfer takes place on the same time scale with the disappearance of dimerization. When the pulse is weak and long, the dimerization-induced ferroelectric polarization is disordered first to restore the inversion symmetry, and then the charge transfer takes place to bring the system to a neutral state. The difference between the two time scales increases as the pulse is weakened. Between these time scales, a paraelectric ionic phase is realized. In the case of intramolecular excitations at 1.55 eV, such a new ionic phase with disordered polarizations is suggested to appear by time-resolved X-ray diffraction. 23) This similarity may be due to the fact that, when the pulse is weak, the supplied energy is not directly used to transfer charge density along the chain. It is furthermore shown theoretically that infrared light polarized along the chain can also induce the ionic-toneutral transition if the excitation density exceeds a critical value, which is currently beyond the value experimentally achieved. When the neutral phase is photoexcited, the linear behavior is observed by plotting the final ionicity as a function of the increment of the total energy (Fig. 7). 24) The neutral-to-ionic transition proceeds in an uncooperative manner. If the oscillating electric field is turned off before the transition is completed, the ionicity remains intermediate unless the energy dissipation is taken into account. The growth of a metastable domain is not spontaneous but forced by the external field. The final state is determined merely by how many photons are absorbed to increase the ionicity. This result is consistent with the experimental finding in the neutral-to-ionic transition induced by intrachain charge-transfer photoexcitations. 21) By intramolecular excitations, on the other hand, a transition from the neutral phase to the ionic phase with three-dimensionally ordered ferroelectric polarizations is shown to be induced, as clearly demonstrated by X-ray diffraction. 25) The reason for this difference is unclear yet. The qualitative difference described above between the two transitions is not due to the different inversion symmetry between the dimerized ionic and the regular neutral states because the difference survives even if the electron-lattice coupling is turned off so that both ionic and neutral states have even parity with respect to the space inversion. The most plausible reason for the difference is the fact that the ionic state is a Mott insulator caused by the electron correlation, whereas the neutral state is a band insulator caused by the band structure. In a Mott insulator, all the electrons are so correlated that any one of them cannot easily make the first move. However, above the threshold excitation density, they cannot tolerate the increased total energy. Once some electrons move, they trigger the collective motion. One electron is transferred after another like dominoes. The collective dynamics in the ionic phase are due to the electron-electron interaction that realizes the Mott insulator phase. They cannot be described by stochastic approaches of refs. 4 and 5. In a band insulator, electrons move individually. The supplied energy is merely consumed to transfer electrons from the donor to acceptor molecules almost independently. They do not strongly influence the motion of other electrons. Thus far, we have focused on the dynamics in purely one-dimensional systems. Although the short-time behavior may not suffer from interchain interactions, they are important for the condition of coherence, which can be manifested by a double pulse. 26) The coherent motion of the macroscopic neutral-ionic domain boundary 18) is possible only when interchain interactions are strong enough. Although the effect of interchain elastic couplings is considered in the context of confinement of a metastable domain, 17) interchain electronelectron interactions are much stronger. The intramolecular charge distribution in the TTF and CA molecules is calculated by the ab initio quantum chemical method. 27) The interchain electrostatic energies between neighboring molecules are smaller than but comparable to the intrachain ones. Because the molecules are tilted, the relative repulsion strengths are very different from the naive expectations based on the intermolecular distance measured in terms of the center of mass of each molecule. In the ionic phase, the interchain attractive interaction between the neighboring donor and acceptor molecules along the b ′ axis is larger than any interchain repulsive one between the donors or between the acceptors. The consequent electrostriction is theoretically shown 28) to be responsible for the pressure-temperature phase diagram of the TTF-CA complex containing the paraelectric ionic phase 29) and for the discontinuous contraction along the b axis. 30) Then, we add to the model (10) with (14) interchain electron-electron interactions, 31) H el inter = l,j (U ⊥ δn l,j δn l,j+1 + V ⊥1 δn l,j+1 δn l+1,j ) , where j is the chain index, and the donor-acceptor coupling strength V ⊥1 is slightly larger than the donor-donor or acceptor-acceptor one U ⊥ . 27) When the ionic phase is photoexcited, the transition dynamics depend on the strengths of the interchain couplings. With weak interchain couplings, for instance, halves of the values for TTF-CA, the interchain correlation is very weak during the transition. A first neutral domain is easily created by a low density of photons. If the electric field is switched off immediately after this absorption, the residual chains remain ionic. By continuing the application of the electric field, next domains are created in the neighboring chains. Many photons are finally needed to complete the transition into the neutral phase. With strong interchain couplings comparable to those in TTF-CA, the interchain correlation is strong during the transition. To create a first neutral domain, more photons need to be absorbed. Once a domain is nucleated, nearby domains are almost simultaneously created in the neighboring chains to grow cooperatively. Their growth cannot be stopped even if the electric field is switched off immediately after the appearance of the first domain. As a consequence, the density of absorbed photons needed to complete the transition is lower than in the weak-coupling case. Because of the strong interchain correlation, the growth of metastable domains coherently proceeds, which is consistent with the experimentally observed, coherent motion of the macroscopic neutral-ionic domain boundary. 18) Photoinduced phases in MX and MMX chains Besides the TTF-CA complex described above, many other molecular materials have been or are currently studied to show their own transition dynamics. Some of them show very asymmetric transitions between two different electronic phases. In a halogen-bridged nickelchain compound, the ultrafast photoinduced Mott transition from a charge-transfer insulator to a metal is observed. 32) Photoinduced optical responses of onedimensional half-filled Hubbard models with and without alternating potentials are studied by the exact diagonalization method to show a large shift of the spectral weight to low energies including the Drude component in the Mott insulator phase, which are similar to the response caused by chemical doping. 33) In halogen-bridged binuclear platinum-chain compounds, R 4 [Pt 2 (pop) 4 I]nH 2 O (pop=P 2 O 5 H 2− 2 ) with cation R, the transition from a charge density wave (CDW) to a charge-polarization (CP) state is photoinduced much more easily than its inverse transition. 34) The electronic phases of these compounds are controlled by the distance between the neighboring binuclear units d MXM both when the counterion R is varied and when the applied pressure is varied. The nonmagnetic CDW phase is observed for small d MXM , while the paramagnetic CP phase for large d MXM . The charge density is disproportioned among binuclear units in the CDW phase, and within each binuclear unit in the CP phase (Fig. 8). In both phases, the iodine ion approaches the platinum ion with less electron density. The dependences of the electronic phase and its physical properties on d MXM are well explained by perturbation theories from the strong-coupling limit as well as the exact diagonalization method. 35 36) In the CDW phase, the charge is transferred to the neighboring unit to form a transient CP domain. In the CP phase, the charge is transferred within the unit to merely reverse the polarization. connecting the degenerate CP phases. The coherence recovery is achieved with much more difficulty in the CDW phase. Concluding Remarks We have summarized the history of theories for photoinduced phase transitions, focusing on the development from stochastic to coherent transition dynamics. Both dynamics are characterized by cooperativity. Coherent dynamics are direct consequences of interactions among electrons and/or between electrons and lattice displacements. Photoinduced transition dynamics between different electronic phases of different materials will be studied further in appropriate models for understanding each mechanism and for exploring new possibilities. Current and future issues to be solved will include differences between photoinduced and thermally induced phases, origins of different nonlinear characters, coherence originating from phonons and/or that from electronic excitations, relative dynamics among spins, charge and lattice, relations to quantum phase transitions or quantum natures of phonons, and how to control the dynamics by tuning the laser pulse.	2019-04-14T02:10:00.633Z	2005-11-05T00:00:00.000	{ "year": 2005, "sha1": "5b5bfa4f12bf181a49086117d10e07275a3b78a9", "oa_license": null, "oa_url": "http://arxiv.org/pdf/cond-mat/0511137", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "4ec491b9811987f4b90fedcf40f33f2d9e37f85f", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
266891413	pes2o/s2orc	v3-fos-license	A zinc metabolism-related gene signature for predicting prognosis and characteristics of breast cancer Background Breast cancer is one of the most serious and prevalent malignancies. Zinc is commonly known to play a crucial role in the development and progression of breast cancer; however, the detailed mechanisms underlying this role are not well understood. This study aimed to develop a zinc metabolism-related gene (ZMRG) signature based on a multi-database study to predict patient prognosis and investigate the relationship between drug therapy response and immune enrichment. Methods Data for breast cancer samples from The Cancer Genome Atlas and Gene Expression Omnibus databases were screened for zinc metabolism-related genes using the Molecular Signature Database. Cox and Least Absolute Shrinkage and Selection Operator regressions were performed to construct a ZMRG signature. To assess the predictive performance of the gene signature, Kaplan–Meier analysis and receiver operating characteristic curves were used. Additionally, we utilised single-sample gene set enrichment analysis, the Tumour Immune Estimation Resource, the Genomics of Drug Sensitivity in Cancer database, and the Cancer Therapeutics Response Portal to investigate the association between the tumour microenvironment and drug sensitivity. Quantitative PCR was used to assess the expression of each gene in the signature in breast cancer cell lines and patient samples. Results Five ZMRGs were identified (ATP7B, BGLAP, P2RX4, SLC39A11, and TH) and a risk profile was constructed for each. Two risk groups, high- and low-risk, were identified in this way, and the high-risk score subgroups were found to have worse prognosis. This risk profile was validated using the GSE42568 dataset. Tumour microenvironment and drug sensitivity analyses showed that the expression of these five ZMRGs was significantly associated with immune response. The high-risk group showed substantial immune cell infiltration and enrichment of immune pathways, and patients were more sensitive to drugs commonly used in breast cancer. Conclusion The ZMRG signature represents a new prognostic predictor for patients with breast cancer, and may also provide new insights into individualised treatment of breast cancer. Introduction Breast cancer (BC), which has been increasing in incidence over the past few decades, is the most prevalent cancer globally and continues to be responsible for a significant number of cancer-related deaths (1).BC can be classified into four molecular subtypes based on estrogen or progesterone receptor expression and Her-2 gene amplification.The different subtypes have different risk profiles and optimal treatment strategies (2).Although the early diagnosis of BC improves survival, a proportion of BC patients still develop metastases after treatment or are diagnosed after metastases have occurred (3).Drug resistance is the central challenge in BC therapy, and it has now been demonstrated that the development of drug resistance is closely linked to the tumour immune microenvironment (TIME) (4). Zinc (Zn) is an essential trace mineral in the human body and plays a vital role in various biological processes.Zn metabolism relies on two proteins, ZIP and ZNT, for intracellular and extracellular transport of Zn ions (5,6).Zn acts as a cofactor for more than 300 enzymes and is involved in cell signalling, proliferation, immune function, oxidative stress, and apoptosis (7).Studies have shown that disruption of Zn metabolism contributes to the development of various cancers, including BC, prostate cancer, pancreatic cancer, and hepatocellular carcinoma (8).Zn has also been shown to be involved in the metastasis of BC (9).Patients with BC often exhibit elevated tumour and reduced serum Zn levels (10).Abnormalities in Zn metabolism may directly contribute to BC development and progression.However, the detailed molecular mechanisms underlying the role of Zn in the promotion of tumourigenesis and metastasis requires further exploration.Understanding these mechanisms may provide new insights for breast cancer treatment. In this study, we aimed to investigate the relationship between Zn and the prognosis and characteristics of BC.We screened RNA expression data from normal and BC tissue samples for differentially expressed Zn metabolism-related genes (ZMRGs) and constructed a prognostic signature based on these ZMRGs.We further analysed the relationship between ZMRGs and the TIME and performed relevant drug sensitivity analysis. Data acquisition and collection RNA sequencing data and clinical information for BC and normal cases (111 normal and 1033 BC cases) were acquired from The Cancer Genome Atlas (TCGA) data portal (https:// portal.gdc.cancer.gov/).The detailed clinical data collected included age, survival time, status, and staging.GSE42568 was used as a validation set and was acquired from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/).ZMRGs were extracted from the Molecular Signatures Database (MSigDB) (https:// www.gsea-msigdb.org/gsea/msigdb/). ZMRG prognostic signature construction and validation Differentially expressed genes (DEGs) between BC and normal tissues were identified using the "limma" R package.DEGs were identified under the criteria \|log2-fold change (FC)\| >0.5 and P < 0.05 after adjusting for false discovery rate (FDR).Protein-protein interaction (PPI) information for the 76 ZMRGs was obtained from the STRING database (http://www.string-db.org/,version 11.5).We identified the ZMRGs that were significantly associated with prognosis using univariate analysis.Next, we selected stable prognostic genes (R package "glmnet") using Least Absolute Shrinkage and Selection Operator (LASSO) regression.We constructed a prognostic model from the prognostic ZMRGs using multivariate Cox regression analysis and calculated Zn metabolismrelated risk scores (ZMRS) based on the corresponding coefficients.The ZMRG formula was calculated as "Riskscore = gen1 * coef1 + gen2 * coef2 + gen3 * coef3… + genN * coefN" as a specific expression.Samples were divided into high-and low-risk groups based on the median ZMRS score.Time-dependent curve analysis was generated using the "timeROC" R package to assess the predictive accuracy of the prognostic ZMRGs. Construction and evaluation of the nomogram Univariate and multivariate Cox regression analyses were conducted to identify the independent prognostic values of the risk scores and of several clinicopathological features (including age, staging, and survival time and status).A prognostic nomogram combining the risk score and the clinical characteristics was constructed using the R package "rms". Analysis of correlation with immune status The enrichment scores of 13 immune-related pathways and 16 immune cell subpopulations were compared between the low-and high-risk groups by single-sample gene set enrichment analysis (ssGSEA) using the R package "gsva".The Tumour Immune Estimation Resource (TIMER2.0,https://cistrome.shinyapps.io/timer/) was used to obtain information on the expression of ZMRGs and immune cell infiltration.We analysed the relationship between ZMRG expression and the abundance of six types of immune cells: B cells, CD4 + T cells, CD8 + T cells, neutrophils, macrophages, and dendritic cells (DCs). Analysis of chemotherapeutic sensitivity and potential drugs in BC To explore whether the risk prognostic signature is associated with chemotherapy resistance in BC, we used the R package "pRRophetic" to predict the half-maximal inhibitory concentration (IC 50 ) of chemotherapeutic agents.Two databases were consulted for potential therapeutic drug screening: the Genomics of Drug Sensitivity in Cancer (GDSC, https:// www.cancerrxgene.org/) and the Cancer Therapeutics Response Portal (CTRP, http://portals.broadinstitute.org/ctrp/).From the GDSC database, the IC 50 and corresponding mRNA gene expression for 265 small molecules from 860 cell lines was collected.This same information was collected for 481 small molecules from 1001 cell lines from the CTRP database.Both datasets were analysed using Pearson correlation analysis to identify the correlation between target gene mRNA expression and drug IC50. Quantitative RT-PCR in cell lines and breast tissues BC and paracancerous tissues from 14 patients with BC were sourced from the First Hospital of Jilin University.Tissue collection was approved by the hospital's ethics committee.RNA was extracted from the cell lines and breast tissue and was reversetranscribed to cDNA.Real-time PCR was then performed using the 2 × RealStar Fast SYBR qPCR Mix (GeneStar, Beijing, China) to quantify the transcripts of the five ZMRGs identified as prognostic signatures.Primers were purchased from Jilin Comate Bioscience Co., Ltd.(Jilin, China).The ATP7B, BGLAP, P2RX4, SLC39A11, TH, and GAPDH oligonucleotide primer sequences were: ATP7B Target gene expression was quantified relative to the internal control, GAPDH. Statistical analysis All statistical analyses were performed using R software (version 4.3.0).The Log-Rank test was used to detect survival rates in each group, the Wilcoxon test was used to compare the differences between the two data groups, and the Kaplan-Meier method was used to plot the survival curves of patients in both groups.Univariate Cox regression and multivariate Cox analysis were used to construct the prognostic signatures and assess their value.P < 0.05 was considered statistically significant. Identification of ZMRGs and DEGs in BC The workflow diagram of this study is shown in Figure 1.The age and tumour stage differences across the cases analysed in this study are shown in Figures 2A, B. To pinpoint genes related to Zn metabolism, we chose multiple pathways using the MSigDB database and identified 76 related genes (Figure 2C).To further explore the interactions of the selected ZMRGs, we performed a PPI analysis (Figure 2D).We set the minimum interaction score required for the PPI analysis to 0.9 (highest confidence level).The correlation network containing all the ZMRGs is shown in Figure 2E.In addition, we found 29 DEGs by comparing gene expression patterns in 1033 BC samples with those in 111 normal breast samples using thresholds \|log2FC\| > 0.5 and FDR < 0.05.Among these DEGs, 19 were upregulated and 10 were downregulated in BC (Figure 2F).The location on the chromosome and the expression levels of the 29 DEGs are illustrated in a circular plot (Figure 2G).The features of association between these genes were revealed using correlation matrix plots (Figure 2H).Workflow diagram of this study. Development of a ZMRG prognostic signature Five significant prognostic ZMRGs were identified using univariate Cox regression in the TCGA-BC cohort (Figure 3A).All 29 DEGs were screened for prognostic value using LASSO regression (Figures 3B, C).Subsequently, the corresponding coefficient values were extracted and individual risk scores were calculated based on the coefficient-weighted expression levels of the selected genes.The risk score was calculated as follows: risk score = (-0.052542× ATP7B expression) + (-0.436158 × BGLAP expression) + (-0.065372 × P2RX4 expression) + (0.017788 × SLC39A11 expression) + (0.223976 × TH expression).Patients were stratified into high-and low-risk groups according to the median cut-off values (Figure 3D).Principal component analysis (PCA) demonstrated that samples were clearly divided into two groups according to their risk scores (Figure 3E).The distributions of the risk scores and survival metrics are shown in Figure 3F.With the increase in risk score, the risk of death increased, while the time to overall survival (OS) decreased.Kaplan-Meier survival curve analysis revealed that patients in the high-risk group had significantly worse OS than those in the low-risk group (P < 0.001; Figure 3G).Time-dependent ROC curve analysis was used to estimate the predictive power of the prognostic ZMRGs (Figure 3H).The AUCs were 0.613 and 0.644 for the ZMRGs at 1 and 3 years of follow-up, respectively, indicating relatively good predictive accuracy. The signature significantly predicted survival in univariate and multivariate Cox regression analyses.Univariate Cox regression analysis indicated that the risk score was an independent predictor of poor survival in patients with BC (HR = 1.8394, 95% CI: 1.2773-2.6488; Figure 4A).Multifactorial analysis yielded the same result, with the risk score being identified as an independent prognostic factor (HR = 2.2868, 95% CI: 1.5703-3.3303; Figure 4B) after adjusting for other clinical characteristics, including age and tumour stage.In Frontiers in Immunology frontiersin.orgaddition, we generated a heatmap of the clinical characteristics of TCGA cohort (Figure 4C). Validation of the ZMRGs signature in GEO dataset To validate the ability of the selected ZMRG signature to predict BC prognosis, we selected the external cohort GSE42568 (104 BC samples) from the GEO database.The risk model estimated risk scores for selected patients were calculated using the same formula and the same cut-off values were used to classify them as high-or low-risk patients (Figure 5A).Similar to the results obtained from the original TCGA cohort, patients in the high-risk group were more likely to die (Figure 5B).Survival analysis revealed significantly different survival rates between the two risk groups (Figure 5C).The AUCs were 0.808 and 0.758 for the ZMRGs at 1 and 3 years of follow-up, respectively, meaning we also achieved good predictive accuracy in the GEO cohort (Figure 5D). Evaluation of a clinicopathologic nomogram We constructed a nomogram by combining the ZMRG signature with two clinical characteristics (age and AJCC stage) to better predict 1-, 3-, and 5-year OS in patients with BC (Figure 6A).The plotted calibration curves showed a good agreement between the predicted and actual survival curves, confirming that the column line graphs had satisfactory predictive discrimination (Figure 6B). Association between immune status/ tumour immune infiltration and risk score of ZMRGs Zn metabolism plays a significant role in breast tumour immunity.To further investigate the association between the ZMRG risk scores and tumour immunity, the enrichment scores of 16 immune cell subpopulations and 13 immune-related pathways were compared between the low-and high-risk groups singlesample gene set enrichment analysis (ssGSEA).The results showed that eight immune cell types (DCs, macrophages, mast cells, NK cells, Tfhs, Th1 cells, Th2 cells, and Tregs) were significantly more enriched (P < 0.05) in the high-risk group than the low-risk group, meaning higher levels of immune cell infiltration were present in the high-risk group (Figure 7A).In terms of immune-related pathways, nine pathways showed significantly higher (P < 0.05) immune activity in the high-risk group than in the low-risk group (Figure 7B).Next, the correlation between the expression of the five prognostic ZMRGs (ATP7B, BGLAP, P2RX4, SLC39A11, and TH) and the level of BC immune infiltration was explored.The results suggested that the expression of these ZMRGs was associated with infiltration patterns of many immune cells.The expression level of ATP7B was positively correlated with the level of immune infiltration of CD8 + T cells (P = 1.58 × 10 -2 ), CD4 + T cells (P = 3.71 × 10 -2 ), and macrophages (P = 7.45 × 10 -9 ) (Figure 7C).The expression level of BGLAP was positively correlated with the level of immune infiltration of B cells (P = 3.15 × 10 -4 ), CD8 + T cells (P = 5.36 × 10 -15 ), macrophages Validation of risk characteristics in the GEO cohort using ZMRG risk scores.(P = 3.98 × 10 -8 ), neutrophils (P = 2.30 × 10 -5 and dendritic cells (P = 4.05 × 10 -3 ) were negatively correlated with the level of immune infiltration (Figure 7D).The level of P2RX4 expression positively correlated with the level of immune infiltration of macrophages (P = 5.38 × 10 -3 ) and negatively correlated with the level of immune infiltration of dendritic cells (P = 3.67 × 10 -2 ) (Figure 7E).The expression level of SLC39A11 was positively correlated with the level of immune infiltration of CD8+ T cells (P = 3.89 × 10 -2 ) and macrophages (P = 1.12 × 10 -6 ) (Figure 7F).The expression level of TH was positively correlated with the level of immune infiltration of CD8 + T cells (P = 7.73 × 10 -3 ) and was negatively correlated with the level of infiltration of macrophages (P = 4.25 × 10 -2 ) (Figure 7G).To further validate the role of ZMRGs in the biological process of immunity, we performed a GO enrichment analysis of DEGs between the high-risk and low-risk groups.The results showed that ZMRGs were significantly enriched in the biological processes of immunization, consistent with the results of this part of the study (Supplementary Figure 1). Exploration of chemotherapeutic response and potential drugs for BC based on ZMRGs To explore the value of the ZMRG-related risk scores for predicting common chemotherapeutic agents for BC, we used the "pRRophetic" database to predict the chemotherapeutic responses in BC patients.We calculated the IC 50 values of commonly used drugs in the high-and low-risk groups.The IC 50 values of drugs recommended in the current major BC treatment guidelines include cisplatin, docetaxel, doxorubicin, vinorelbine, palbociclib (PD.0332991), and olaparib (AZD2281), among others, were lower in the high-risk group than in the low-risk group, meaning there was greater sensitivity to these drugs in the high-risk group (Figure 8A).We also identified several potential BC therapeutics using the GDSC and CTRP databases.According to the GDSC database analysis, BX-795, bleomycin, and others may be effective drugs (Figure 8B).Based on analysis of the CTRP database, CHIR-99021, fluvastatin, lovastatin, and others may be effective (Figure 8C). Validation of the mRNA expression level of ZMRGs The expression of the ZMRGs which form the signature was validated using qRT-PCR.We detected gene expression in several common breast cancer cell lines and a normal breast epithelial cell line, as well as comparing expression between breast cancer and paracancerous tissues from 14 patients.The results showed that P2RX4 and SLC39A11 were highly expressed in most breast cancer cell lines (Figures 9A, B), and ATP7B and P2RX4 were significantly up-regulated in BC tissues (P < 0.05) (Figures 9C, D), which is consistent with the results of previous bioinformatics analyses based on the TCGA database and other results described in the literature.Comparison of ssGSEA scores between the two risk groups and correlation between the expression of the five ZMRGs and immune infiltration in BC. Discussion In recent years, the relationship between BC and Zn has received increasing attention.Zn is an important trace element in the body and is involved in regulating numerous metabolic processes.Abnormal Zn metabolism can lead to the development and progression of BC (11).However, studies on the prognostic value of genes related to Zn metabolism in BC have not yet been reported.Prognostic signatures are essential for precision medicine and can help guide clinical decisions.Therefore, we explored the ability of ZMRGs to predict prognosis and established a prognostic signature to provide more accurate therapeutic decisions and diagnostic approaches for BC. Our prognostic signature included five ZMRGs (ATP7B, BGLAP, P2RX4, SLC39A11, and TH).Previous studies have explored the relationship between ZMRGs and tumours, with all ZMGRGs studied being found to play a role in tumour development and progression.ATP7B is highly expressed in many tumours including BC, ovarian cancer, oesophageal cancer, gastric cancer, and hepatocellular carcinoma (12,13).ATP7B is not only involved in tumour progression, but also in the development of platinum drug resistance (14)(15)(16).A previous study reported that BGLAP was associated with advanced breast cancer staging (17).P2RX4 belongs to the P2 purinergic receptor family, which is commonly upregulated in a variety of tumours and is associated with poor prognosis (18).Specifically, P2RX4 plays a key role as a regulator in inflammation and immune cell function (19).A previous study showed that P2RX4 was upregulated in BC samples, which is largely similar to our findings.It has also been demonstrated that P2RX4 enhances BC invasiveness, breast tumour growth, and metastasis (20).Given its role in promoting tumour formation, P2RX4 has emerged as a potential therapeutic target (21).SLC39A11 is a solute carrier of membrane transport protein family 39 (SLC39A).SLC39A11 was previously found to be significantly upregulated in BC tissues compared to normal breast tissues (22), which is similar to our findings.In colon cancer, SLC39A11 expression is also upregulated, possibly in response to the increased Zn demand in cancer cells (23).SLC39A11 is also a prognostic indicator and therapeutic target in lung adenocarcinoma (24).TH encodes a protein that is involved in the conversion of tyrosine to dopamine.Higher levels of TH expression are associated with worse outcomes, and positive TH expression is a poor prognostic indicator of metastatic neuroblastoma (25). It is well known that Zn is important for immune function and that abnormalities in Zn metabolism affect immune cells and lead to alterations in host defences (26).Zn is involved in the regulation of multiple intracellular signalling pathways in innate and adaptive immune cells (27,28).In innate immunity, Zn deficiency affects the activity of natural killer cells and phagocytosis by macrophages (8,29).In adaptive immunity, some Zn transporter proteins that are highly expressed in T cells, such as Zip6 and Zip8, are directly involved in T cell activation via Zn endocytosis (30).Additionally, Zn is an essential cofactor that directly affects the biological activity of thymosin, which in turn affects T-cell generation, maturation, differentiation, proliferation, and functional changes (31,32).Our ssGSEA results confirmed the presence of large immune cell infiltration and enrichment of immune pathways in the high-risk group, suggesting that more immune pathways were activated in the high-risk group than in the low-risk group.These results suggest that the high-risk group is generally immunologically activated, indicating that patients with a high-risk signature are more likely to show a better response to immunotherapy, which may be a focus for future therapeutic development. Interactions between different components of the tumour microenvironment affect tumour development and metastasis and promote BC chemoresistance (33).Chemoresistance during BC treatment significantly affects prognosis.The mechanism of chemoresistance in tumours is complex and involves interactions between numerous signalling pathways (34).Clarifying the mechanisms underlying drug resistance is challenging.Therefore, it is crucial to accurately differentiate drug-sensitive populations.In our study, drug sensitivity analysis was performed on high-and low-risk groups, defined using our ZMRG signature, drawn from the GDSC and CTRP databases.High-risk patients were found to be more sensitive to drugs commonly used in BC, such as cisplatin, docetaxel, doxorubicin, vinorelbine, palbociclib, and olaparib.Using our ZMRG prediction model to evaluate the sensitivity of BC drug therapy, we hope to provide new therapeutic ideas for BC treatment and more precise and individualised treatments for patients with BC in future clinical practice. Despite improving confidence in our results through our multifaceted analyses, database validation, and validation of multiple BC cell lines and patient samples, this study has some limitations.First, this was a retrospective study, and data for all patients with BC were collected from public databases.Second, further validation of our findings is needed in future studies because the drug sensitivity prediction conducted here is based only on bioinformatics analyses.In addition, the molecular mechanisms underlying the function of the ZMRGs which contribute to our risk score in BC need to be explored in more depth. In summary, we constructed a prognostic signature based on ZMRGs to predict the prognosis of BC patients.This model is related to the immune microenvironment and can distinguish the immune status of patients with BC.More importantly, it could be used to provide new treatment options for BC patients in different risk groups. 2 FIGURE 2 Expression and Genetic characteristics of ZMRGs in TCGA-BRCA cohort.(A) Age distribution in the TCGA-BRCA cohort.(B) Tumour stage distribution in the TCGA-BRCA cohort.(C) Heatmap of 76 differentially expressed ZMRGs. (D) PPI analysis among the 76 ZMRGs.(E) The correlation network of the ZMRGs.(F) Volcano plot illustration of differentially expressed ZMRGs in the TCGA-BRCA cohort.(G) The location of the ZMRGs on chromosomes.(H) Correlation matrix among the ZMRGs. 3 FIGURE 3 Correlation analysis between prognosis and characteristics of BC patients from the TCGA-BRCA database.(A) Univariate Cox regression of the five ZMRGs.(B) LASSO regression of the five ZMRGs.(C) Cross-validation in the LASSO regression.(D).The distribution of risk scores of the five ZMRGs.(E) The PCA plots based on risk scores.(F) The distribution of OS status of the five ZMRGs.(G) Kaplan-Meier survival curves for OS in the two risk groups.(H) AUC values of ROC curves for risk scores. 4 FIGURE 4 Prognostic value of the risk signature.(A) Univariate Cox analysis on OS for the risk score in TCGA-BRCA cohort.(B) Multivariate Cox regression on OS for the risk score in TCGA-BRCA cohort.(C) Heatmap of the association between clinical and pathological features in the two risk groups. FIGURE 6 FIGURE 6 Development of the prognostic nomogram.(A) Nomogram based on the five ZMRGs comprising the signature and the clinical information.(B) The calibration plots for predicting 1-, 3-, and 5-year survival probabilities. 8 9 FIGURE 8 Potential therapeutic agents for BC based on ZMRG risk scores.(A) Comparison of IC 50 values commonly used drugs recommended in BC treatment guidelines in two risk groups.(B) Drug sensitivity analysis based on the GDSC database, drugs such as BX-795 and bleomycin may be therapeutically effective.(C) Drug sensitivity analysis based on the CTRP database, drugs such as CHIR-99021, fluvastatin, and lovastatin may be therapeutically effective.Positive correlations are represented by red bubbles and negative correlations by blue bubbles.The darker the colour of the bubbles the higher the correlation.The size of the bubbles is positively correlated with the significance of FDR.A black border on the bubbles indicates FDR ≤ 0.05.	2024-01-10T16:17:21.081Z	2024-01-08T00:00:00.000	{ "year": 2024, "sha1": "e81c4f21de12b047ff2adb3afba222092087a2f8", "oa_license": "CCBY", "oa_url": "https://www.frontiersin.org/articles/10.3389/fimmu.2023.1276280/pdf?isPublishedV2=False", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "10d990eaad576b40317556961d4bc5dae23736c3", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
148576210	pes2o/s2orc	v3-fos-license	Restructuring and Resisting Education Reforms in Chicago’s Public Schools education policy analysis The Landscape of Education “Reform” in Chicago: Neoliberalism Meets a Grassroots Movement : This article examines the dialectics of Chicago’s neoliberal education policies and the grassroots resistance that parents, teachers, and students have mounted against them. Grounding the analysis in racial capitalism and neoliberal urban restructuring, I discuss interconnections between neoliberal urban policy, racism, and education to clarify what is at stake for communities resisting Chicago’s policies. The paper describes deep and pervasive racial inequities, school closings, privatization, and disenfranchisement driving organized opposition and the labor-community alliance at the center of organized resistance. I argue that neoliberal education policy is racialized state violence, and education is a battleground for racial justice and Black self-determination. I conclude with observations on Chicago’s experience so far that might be useful in other contexts. link education and racial justice. 2 In Chicago, African American and Latinx parents and community organizations took the lead, beginning in 2004. Their organizing attracted teachers who, in 2008, formed a radical caucus of the Chicago Teachers Union (CTU). Just two years later, the caucus won the leadership of the third largest teachers union in the US. The new leadership's explicit opposition to neoliberal policies, its alliance with parents and community members, and its social movement organizing signaled the rebirth of teacher unionism in the US on new grounds (Lipman & Gutstein, 2013). Relentless protest by a multi-racial coalition of parents and a revitalized CTU have made Chicago a focal point of national resistance. But I argue in this paper that the struggle over education in Chicago is about more than schools. It is about race and capital, state violence, claims to urban space, and political power. I argue that resistance to education policy in Chicago is an expression of the rejection of neoliberal policies to remake the city for capital accumulation (Brenner & Theodore, 20002;Harvey, 2005) and to simply abandon, contain, or drive out those whose lives do not matter and are, in fact, disposable in a context of corporate hegemony, elite consumption, and whiteness (Haymes, 1995;Shabazz, 2016;Wilson, 2007). Chicago is, "Two cities really…one white, one black. One for the rich, one for the poor." The dialectic of economic, social and spatialized inequalities and resistance to them is rooted in the city's historical amalgam of racial segregation and machine politics, collusion of the state and capital for real estate profiteering and disinvestment in Black communities (Coates, 2014), and the unfinished business of Chicago's Black freedom movement (Danns, 2014;Homel, 1984). Chicago's education landscape is integral to this dialectic. There is a robust critical literature on neoliberal education policy (e.g., Au & Ferare, 2015;Ball, 2012;Hursh, 2015;Saltman, 2015), its racialized agendas and effects (Dumas, 2013;Picower & Mayorga, 2015;Scott, 2011), and its imbrication with urban development and political economy (Buras et al, 2010;Lipman, 2011Lipman, , 2016Pedroni, 2011). Extending this scholarship, in this article, I aim to contribute to a deeper understanding of the dialectics of neoliberal urban education "reform" and resistance as an expression of resistance to racial capitalism. I examine the stakes in the battle over education and what can be learned from Chicago's labor-community alliance and grassroots movement. I begin by grounding the analysis in racial capitalism and neoliberal restructuring of urban space. I describe my methodology and data and provide a brief overview of Chicago as a laboratory for neoliberal restructuring of public education and resistance. In the next section, I summarize the deep and pervasive racial inequities in CPS as a root of resistance. Then I discuss interconnections between neoliberal urban policy, racism, and education to clarify what is at stake for communities resisting Chicago's policies. In this section, I focus on education policy as racialized state violence and propose that education is a battleground for racial justice. In my discussion of resistance, I foreground the formation of a community-labor alliance, the strategic role of the Chicago Teachers Union and centrality of Black and Latinx parents and the struggle for racial justice. I conclude with some observations on Chicago's experience so far that might be useful in other contexts. Racial Capitalism and Urban Space Renewed interest in Cedric Robinson's (2000) analysis of racial capitalism and recent scholarship (Camp, 2016;Childs, 2016;Melamed, 2011;Taylor, 2016) have invigorated discussion of the co-constitution of capitalism and racism in the US. From the dawn of capitalism, white supremacy was mobilized for the accumulation of capital, launching the long-term historical structures and ideologies that made capitalism a world system and that continue to shape the economy and social life today (Robinson, 2000;Williams, 1944Williams, /1994. The centrality of white supremacy to neoliberal accumulation strategies is the present-day iteration of this dynamic. Neoliberal social policies and methods of governance that enable capital accumulation are created, instituted, and organized through the cumulative effects of past discrimination and structural racism (Taylor, 2016). For example, privatization of public housing and schools in African American and Latinx communities is enabled by historical racially discriminatory mortgage lending, and disinvestment in public housing and schools and is justified by racist ideologies (Coates, 2016;Imbroscio, 2008;Lipman, 2009;Pedroni, 2011). George Lipsitz (2014) describes this process: The ideology of contemporary capitalist neoliberalism needs to disavow the idea of race, because race references historical social identities not reducible to market relations… [yet] neoliberalism needs race even more than previous stages of capitalism did. By making public spaces and public institutions synonymous with communities of color, neoliberals seek to taint them in the eyes of white workingclass and middle-class people, who then become more receptive to privatization schemes that undermine their own stakes in the shared social communities that neoliberalism attempts to eliminate (p. 11). Conversely, capitalism and its crises have shaped the character of white supremacy and the racial order in the US over the past three centuries (Marable, 2015(Marable, /1983. Evolving strategies of capital accumulation have shaped racial discourses and how racism is structured, experienced, and legitimated, e.g, from legal segregation and Black Codes to civil rights to liberal multiculturalism and colorblind racism (Melamed, 2011). Today, racism has taken a neoliberal turn. "Post-racial" colorblind discourse and the ideology of individual responsibility obscure the actual intensification of racism and negate claims of racial injustice while the demonization of people of color justifies dismantling public institutions and the social state. Naming racial capitalism illuminates the social forces driving neoliberal education policies and why racial justice is central to resistance. Neoliberals promote the market as the path to civil rights and racial equality (Melamed, 2011). Charter schools and school "choice" are offered up as the civil rights movement of our time while market logics underpin No Child Left Behind and Obama's Race to the Top federal policies (Lipman, 2016). At the same time, the neoliberal state enforces the racial order through coercive policing and containment (Camp & Heatherton, 2016;Taylor, 2016) and social policies that strip workers, women, and communities of color of rights and public resources, e.g, school closings and punitive racially discriminatory discipline policies (the "school to prison pipeline"). Despite the state's reliance on winning consent for dominant agendas, racialized state violence has always been an indispensable tool of capitalism (Childs, 2016;Robinson, 2000). The explosion of the prison population in the US over the past 30 years and police murders of African Americans are just the most visible ways in which the neoliberal state uses coercion to police the harsh inequalities of neoliberalism and to contain resistance movements (Camp, 2016;Camp & Heatherton, 2016;Taylor, 2016). I will argue that Chicago exemplifies urban education policy as state violence. Cities are laboratories for racialized neoliberal experimentation. Brenner and Theodore (2002) describe neoliberalism as a "complex, multifaceted project of socio-spatial transformation" (p. 16). Neoliberal urban restructuring is accomplished by rolling back welfare-oriented public institutions and social policies and rolling out new institutional arrangements and policies that support capital accumulation through markets, urban entrepreneurism, and speculative investment. Privatization of schools and education services is part of this strategy. The neoliberal city is an entrepreneurial city. Neoliberals mobilize urban space for market-oriented economic growth, international tourism, and elite consumption practices (Harvey, 2001) facilitated by racial exclusion and containment (Lipsitz, 2007;Shabazz 2016). Investment in real estate and the revenue generated by real estate sales and taxes is pivotal in the neoliberal urban economy (Smith, 1996). We see this in national and local policy to dismantle public housing while providing municipal financing of marketoriented projects and subsidies and infrastructure for real estate developers and gentrification (Hackworth, 2007). This is a material and symbolic process of "reclaiming" the city as a space of corporate culture and whiteness, materially and discursively (Hackworth; Haymes, 1995;Smith) while criminalizing low-income people of color and their neighborhoods as "bad" and needing to be "cleaned out" (Camp & Heatherton, 2016). The result is a city of luxury zones and hardship zonesgentrification and displacement, downtown development and neighborhood disinvestment, wealth and poverty, inclusion and exclusion-demarcated by race, ethnicity, and class (Nolan, 2015). Methodology This paper is particularly informed by my participation in education organizing in Chicago over the past two decades as a member of Teachers for Social Justice and the Grassroots Education Movement. I write as researcher and participant in parent-student-teacher-union coalitions and campaigns to defend and transform public education, to contest school closings and privatization and to develop and press for alternatives. Rejecting the binary of rigorous research and political involvement, I claim the "situated objectivity" of engagement in the dialectics I analyze (Gutierrez & Lipman, 2016;Hale, 2008). I have worked alongside parents and teachers and students and benefitted from their insights, and I have seen first-hand the lived experience of CPS policies. While I look carefully at school district data and policies, I privilege the perspectives of the people directly affected, Black and Latinx families, students, and their teachers who bear the brunt of the policies I discuss. Their voices are central to analysis of the dialectics of racial neoliberal policy and community resistance. My discussion also draws on data collected over the past 20 years of following Chicago's education and urban policies and urban education policy nationally. The data include: observations of scores of public meetings; interviews with parents, community members, teachers, and students; and research projects with parents, community organizations, and union researchers. 3 Collaborating with parents, teachers, urban studies scholars, and teacher union researchers, we mapped the relationship of school policies onto housing and neighborhood patterns, community conditions, and demographic factors, and gathered testimonies of people directly affected by CPS policies. We catalogued school disinvestment and tracked CPS policies. The paper also draws on analysis of news media, policy documents, reports, and publicly available demographic and education data. It leans heavily on reports I co-authored on school closings; the intersection of education policies, housing, and urban development; and Chicago's appointed school board (Fleming, et al, 2009;Greenlee, et al, 2008;Gutierrez & Lipman, 2012;Lipman, Vaughan. & Gutierrez, 2014;Lipman, Person & KOCO, 2007;Lipman, Smith & Gutstein, 2012). The data and analyses are a collective product of teachers, parents, community activists, and researchers working together. A Laboratory for Neoliberal Education Policy and Resistance 3 Much of this collaborative work was done through the Collaborative for Equity and Justice in Education at University of Illinois at Chicago (ceje.uic.edu) Chicago has had a long history of racially segregated and inequitable schooling rooted in residential segregation and racist housing and labor markets (Coates, 2016). In 1963, Black parents organized massive school boycotts against overcrowded, racially segregated, under-resourced schools (Danns, 2014). Later in the decade, as the Black Power movement took hold, parents, teachers, and students organized sit-ins, walk-outs, boycotts, and conferences to demand community control of schools, African American history, and more Black teachers and administrators and academic improvements. In 1984, Latinx parents marched on Clemente High School in the largely Puerto Rican Humboldt Park community to demand action on the high Latinx dropout rate and violence in schools (Kyle & Kantowicz, 1992). There were also four teachers' strikes between 1980 and 1987 to contest the state's failure to fund decent salary increases and needed school improvements. During the 1987 strike, a People's Coalition of Black and Latinx parents formed to demand thoroughgoing school reform (Kyle & Kantowicz, 1992). One result was 1988 state legislation establishing elected Local School Councils with parents and community members in the majority. (See Lipman, 2004, for discussion of this history.) Despite gains from these grassroots movements, CPS has persistently failed to provide equitable education for African Americans and Latinx who comprise the vast majority of CPS students (CPS is 90% students of color and 85% low-income), see Danns;Lipman, Gutstein, Gutierrez, & Blanche, 2015;Orfield, 1990). In short, there has never been a "golden age" of education in Chicago. CPS's past failures set the stage for Chicago to be a testing ground for neoliberal education policies. Chicago was one of the first cities to establish mayoral control as a means to fast-track neoliberal policies without interference by elected, publically accountable school boards. In the early years of mayoral control and top-down accountability (1997)(1998), CPS put over 100 schools on probation under direct control of central administration, and retained in grade thousands of mainly Black and Latinx children. The annual posting of standardized test scores was a ritual of public shaming of Black and Latinx students, their teachers and schools, and by implication their families and communities. The purported objectivity of standardized test scores obscured the lived reality of disinvestment and destabilization, of strangled aspirations and futures cut down, of subjectivities disciplined by a culture of public punishment and humiliation (Lipman, 2004). Initially, the discourse of holding schools and teachers accountable for education outcomes and the appeal to reject the "failed policies of the past" resonated with some teachers and parents as a solution to then-Secretary of Education William Bennett's famous pronouncement that Chicago's schools were "the worst in the nation" (Schools in Chicago, 1987). There was little push back from the compliant leadership of the Chicago Teachers Union. But the tyranny of high stakes tests and the degradation of teaching and shaming of schools and students took its toll in teacher demoralization and psychic and educational damage to students (Lipman, 2004). The school district was run like a business with performance management systems and high stakes tests governing teaching. Schools in low-income Black and Latinx communities were subjected to a revolving door of centrally-mandated programs and external monitors and a regimen of testing, scripted curriculum, and punishment for failure (Lipman, 2003). A current of deep discontent with the injustice of it all ran beneath the surface of teacher and administrator compliance, bubbling up in public critiques by parent and education advocacy organizations, teacher and student boycotts of standardized tests in 1999 at several schools, and the exodus of dispirited teachers. In 2004, Mayor Richard M. Daley launched Renaissance 2010 a dramatic plan to make school closings and privatization the centerpiece of CPS's "reform" agenda. Renaissance 2010 was the spark that ignited a grassroots opposition movement. The plan called for closing 60-70 public neighborhood schools and opening 100 new schools, two-thirds as privately-run but publicly-funded charter and contract schools 4 (Lipman & Haines, 2007). The blueprint and rationale were developed by the Commercial Club of Chicago, an organization of the city's most powerful corporate and civic elites (Left Behind, 2003). Hewing to market logics, the Commercial Club contended that "failing" public schools needed to be replaced by a competitive market of private operators who would run schools more efficiently and effectively. Standardized tests provided the metrics for an annual wave of neighborhood school closings and take-overs by private "turn-around specialists." The first target was a plan leaked in fall 2004 to close 20 of the 22 schools in the Midsouth Bronzeville area on Chicago's Black South Side where the Chicago Housing Authority was also dismantling thousands of units of public housing. But parents and community residents organized to stop it, charging that the real goal was "displacement"-to push out Black families and gentrify Bronzeville (Lipman & Haines, 2007). Thus began the battle over school closings and education privatization in Chicago led by coalitions of Black and Latinx community organizations, teacher and parent groups, students, and later the Chicago Teachers Union. Each year the Board of Education voted to close anywhere from 6-18 schools. And year after year, parents and teachers mounted campaigns to save their schools. Despite their problems, these schools were anchors of their communities. Many had proud histories of graduating Black intellectual and cultural icons and teaching multiple generations of the same families. At a community meeting in 2005 to oppose CPS' plan to close Englewood High School, a community member received a rousing ovation when he told a CPS representative, "We're being pushed out of the city under the guise of school reform." A former student said, "When you destroy a community's school, you destroy a community" (author field notes). This sentiment was repeated in school closing hearings around the city. In the North Lawndale community, a parent described school closings as "an act of war on the community" (author field notes). But despite candlelight vigils, pickets, rallies, sleep-outs in front of the Board of Education in winter, press conferences, public testimonials, tears and outrage, and data that countered a narrative of failure, the appointed Board of Education mostly rubber stamped the mayor's annual list of schools to be closed. Simultaneously, the Board steadily authorized more charter schools. CPS also mutated its tactics. When transferring students to schools across neighborhood boundaries led to student conflict and violence, and bad press, CPS added "turn-around schools" to its repertoire of school actions. Students stayed in the building but CPS closed the school and "turned it around" with a private operator and new staff-a closing by another name. When the Board's criteria for "failing schools" was challenged, CPS closed schools for "under enrollment." When there was outrage at the lack of notice before closing a school, CPS began a public hearing process. By 2012, CPS had closed, turned-around, phased out, or consolidated over one hundred schools, directly impacting more 42,000 students, 88% of whom were African American (Vevea, Lutton & Karp, 2013). Schools that were over 99% students of color represented 80% of the affected schools, with Black communities hit hardest (Caref et al, 2012). It was pretty clear that CPS' market agenda was a racial agenda (Weber, Farmer & Donoghue, 2016). And annual authorization of more charter schools illuminated the connection between high stakes testing, school closings, and privatization. Although parents and teachers were only able to stop a handful of closings and turnarounds, through their campaigns Black and Latinx community organizations and parent and teacher groups forged a grassroots coalition. They organized, marched, picketed, slept out, and sat in. They held press conferences and public forums and produced research that made connections between school closings, privatization, gentrification and racism and contested the neoliberal discourse of school choice, educational efficiency, and failing schools. They called for parent and student voice and community-driven school improvement. "Whose schools? Our schools" and "We didn't fail, the system failed us" rang out in marches and rallies. The financial crisis of 2008 provided a warrant to accelerate school closings and institute budget cuts across the city. The budget cuts and the mayor's unpopular unilateral decision to extend the school day brought white middle class families into conflict with CPS, face to face with an imperious school board, and in contact with Black and Latinx parents whose neighborhood schools had been perpetually under-resourced and were threatened with closure. In 2008, teachers in the grassroots coalition launched the Caucus of Rank and File Educators (CORE) to challenge the failure of the leadership of the Chicago Teachers Union (CTU) to fight CPS's neoliberal agenda. In just two years, in a surprise upset, CORE won the leadership of the 30,000 member CTU. These developments set the stage for a multi-racial, multi-class parent/community and labor alliance that would be the backbone of a campaign for an elected school board, for community-driven school transformation, progressive revenue solutions, and the fight for the city itself. In fall 2012, teachers' accumulated anger at serial school closings, coupled with budget cuts and 15 years of high stakes testing and treating schools like businesses exploded in the first teachers' strike in Chicago in 25 years. Under a new militant leadership, the union exercised its power in mass action, shutting down every school, organizing rallies that shut down downtown Chicago, and marching through neighborhoods. The strike was a carnival of resistance, a citywide mass demonstration of the collective power and voice of teachers, parents, and students and a resounding NO to the mayor and the Board. The strike was widely seen as a victory for the union and was celebrated by teachers around the U.S., inspiring new possibilities for teacher unionism in the US (Peterson, 2014). A reinvigorated CTU, under the leadership of CORE, emerged as a central actor in the struggle for public education in Chicago and nationally. Mayor Emanuel's rejoinder was swift and dramatic. Following the strike, he announced a CPS' budget crisis that he claimed would require "right sizing" the district by closing up to 140 "underutilized" schools, a number eventually whittled down to 54. After six months of tumultuous protests, the Board voted to close 50 neighborhood schools, the most ever closed at one time by a U.S. school district. The closings affected 42,952 students (Vevea & Lutton.2013); 79% were African American. To date (2017), CPS has closed, turned around, consolidated, or phased out (all closings by another name) 157 schools. 5 There are now 130 charter schools and 31 schools run by the contract school operator, Academy for Urban School Leadership, most in Black and Latinx communities. Almost one quarter of the school district's 660 schools are privately managed. More than a decade of school closings and charter school openings has changed the education landscape in Chicago-spatially and politically. Some African American areas of the city now have almost no public neighborhood schools left. When CPS closed Dyett High School in 2012, Bronzeville lost its last neighborhood public high school. In 2013, five African American community areas had 60% of children attending schools outside their neighborhood, due largely to school closures (Karp, 2013). However, parents, students and teachers who had never been involved in public action were drawn into organized opposition to the mayor and the Board. Thousands participated in public hearings and organized petitions, rallies, pickets, press conferences, and candlelight vigils, culminating in a three-day march across the city led by a coalition of community, parent, and teacher organizations and the CTU. In the end, the Board closed 50 schools anyway, but the strength of the resistance, the testimonies of children and parents, the potential effects on families and children's safety undermined the mayor's narrative of efficiency and education improvement. Mayor Emanuel got his way, but parents and students won the moral victory. The appointed Board's autocratic decision-making and inability to be held accountable for policies that were devastating to CPS families (Lipman, Vaughan & Gutierrez, 2014) provoked a grassroots campaign for an elected school board that won two city-wide advisory referenda with nearly 90% approval. The referenda put the mayor on the defensive and launched a drive for state legislation mandating an elected representative school board for Chicago. In short, the mayor and Board of Education have paid a high political price for their agenda-a critical loss of legitimacy that raised the stakes, bringing into question their leadership and agenda for the city as a whole. A More Inequitable Education System Opposition to CPS policies is a response to a deeply inequitable education system. Chicago has a well-known history of racially segregated and inequitable schooling, but after 20 years of neoliberal policy, Chicago Public Schools are more unequal by nearly every measure than when mayoral control was instituted in 1995. Unequal opportunities to learn in a racially discriminatory and unjust city (Caref, et al, 2015) have produced persistent racial disparities in test scores and graduation and dropout rates which have not narrowed since 1995 when the mayor took over the school district, and on several measures disparities have increased (Lipman et al., 2015). High stakes testing and accountability, school closings, selective schools and programs, privatization, and CPS budget priorities have exacerbated historical inequalities, creating a more sharply tiered system of learning opportunities and racial disparities (Lipman et al, 2014). A civil rights complaint filed against CPS in 2014 noted that students attending Dyett High School, in the African American Bronzevile community, had no Advanced Placement or honors classes, while students at a mix of five neighborhood and selective enrollment schools had between 5 and 37 advanced classes (Advancement Project, 2014). In 2012, seven times more Black students had neither music nor art classes than students of other races, and, due to charter school proliferation in Black communities, Black students were disproportionately taught by inexperienced first year teachers (Chicago Teachers Union, n.d.). Although the specific data may not be widely known, the inequityis blatant and was a frequent theme in public hearings about school closings. As one parent put it, "If CPS does what it does in schools on the North side [more affluent and white], and gives [us] the same amount of resources, [our school] will get the same results. It is unfair to expect the same result when different resources are being provided" (author field note, quoted in Lipman, Smith, & Gutstein, 2012). Top-down accountability policies and high stakes tests pressed low-scoring neighborhood schools, particularly in Black and Latinx communities, to narrow curriculum to tested subjects (reading and math), replace text books with test prep booklets, focus instruction on test preparation, and mandate scripted instruction in some schools (Hauser, 1999;Lipman, 2004). The schools that CPS put on probation beginning in 1996 were overwhelmingly Black, with an average poverty level of 94% (Parents United for Responsible Education, 1999). Although the Board funneled millions of dollars to outside probation partners to oversee school improvement, by 2013, almost 2.5 times as many schools were on probation as when the policy began-159 elementary schools (more than one-third of CPS elementary and middle schools) and half of the district's high schools (Ahmed-Ullah, 2013). The students and schools experiencing these mandates were almost entirely African American and Latinx. Chicago's education market, known as "options for knowledge," further stratified schooling by race/ethnicity, social class, and neighborhood. In the late 1990s, the Board began expanding a top tier of selective enrollment high schools, mainly in affluent white and gentrifying areas. With competitive admissions based heavily on test scores, students whose elementary schools offered fewer rich learning opportunities clearly had limited access to these (the district's best) high schools (Lipman, 2004). Selective enrollment schools were designed to keep middle class families in the school district and were anchors for real estate development in gentrifying areas (Lipman, 2004;. They benefitted disproportionately from CPS investments, including new or fully renovated buildings, state-of-the-art programs and resources, a rich array of courses and co-curricular opportunities (Williams, 2000), and additional funding from the district's general education budget (Chicago Public Schools, 2014). 6 Those fortunate enough to attend are just 11.7% of CPS high school students (Chicago Sun Times, cited in Lipman et al, 2015) and are almost 2.5 times more white and three times more well off than CPS students as a whole (Novak, 2014). At the same time, CPS opened seven military high schools in low-income African American and Latinx communities and provided additional funds to charter schools and schools managed by the contract operator, Academy of Urban School Leadership (Hood & Ahmed-Ullah, 2012). 7 Budget cuts have also disproportionately impacted neighborhood schools compared with charter schools (Better Government Association, 2015, Karp, 2013. School closings in Black communities and expansion of charter schools also led to loss of Black teachers. From 2000 to 2014, CPS lost 4,385 Black teachers (Illinois Interactive Report Card, 2011), a decline from 40.6% to 23.4% of the teaching force while the percentage of white teachers increased from 45.4% to 62.3 %. Charter schools hire a larger percentage of white teachers than CPS schools (see also Jankov & Caref, in this issue). When CPS launched Renaissance 2010 in 2004, the CPS Board President said CPS could not wait to "rescue" students from failing schools (Lipman & Haines, 2007, p. 475). He promised students would go to "better" schools. Thirteen years and 157 closed schools later, the results have not lived up to the promise. Most students did not end up in schools that performed substantially better on the district's metrics, and in some cases performed worse. Most students displaced from 2004-2006 transferred to schools no better than those that were closed, with 80% transferring to schools in the bottom half of the system on standardized tests (de la Torre & Gwynne, 2009). In the 2013 mass school closings, although 93% of displaced students enrolled in a school with a higher performance rating than their closed school (some only minimally so), only 21% enrolled in schools that could be expected to significantly benefit students academically (de la Torre, Gordon, Moore, & Cowhy, 2015). To counteract powerful protests against mass school closings, CPS promised additional resources in all receiving schools and allocated $83.5 million in additional funds to help students make the transition. However, a CTU report (Caref, Hainds & Jankov, 2014), based on CPS data, found that only 10%, $9.3 million, actually went to the schools for investments in resources. The central office allocated the remaining $74.2 million for moving costs, monitoring closed buildings, and "safe passage programs" to address the many safety concerns raised by the closings. Expansion of charter schools, mainly in Black and Latinx communities, has also not improved education overall. Despite additional public and private funding and greater autonomy from district requirements, their academic performance is mixed and overall no better than traditional, neighborhood public schools (Golab, Schlikerman, & FitzPatrick, 2014;Mihalopoulos, 2014;Institute on Metropolitan Opportunity, 2014). 8 Yet CPS continues to authorize additional charter schools while continuing to cut neighborhood school budgets, even as parents have mounted mass opposition, particularly in Latinx communities. Charter expansion and blatant disparities in resources and concentration of capital improvements in schools disproportionately white and affluent has fueled a deep reservoir of anger and distrust of CPS. Neoliberal Urbanism, Race, and Education Policy Neoliberal education policy is constitutive of the neoliberal restructuring of the city and the politics of race (Lipman, 2011). Situating education in this political-economic dynamic illuminates what is at stake in Chicago's contested education landscape. To borrow from Stuart Hall (1980), race is the "modality" through which neoliberal urban processes-including in education-are experienced, structured, legitimated, and contested. Decades of racial segregation, public and private disinvestment and racially discriminatory housing, employment, and education policy impoverished and destabilized low-income African American (Coates, 2014;Shabazz, 2015) and some Latinx areas in Chicago, and devalued the land and housing stock, making them ripe targets of profit-oriented redevelopment. Over the past 15 years, Chicago dismantled 22,000 units of public housing displacing thousands of primarily Black low-income residents and closed public mental health clinics, early childhood centers, and schools in Black communities. In a race and class conquest of the city (Smith, 1996), real estate developers and investors have remade these areas for upscale housing and retail, subsidized by tax breaks and upgraded infrastructure, e.g., new street grids and streetscapes, parks, and police stations in areas to be gentrified. Both Mayor Daley and Mayor Emanuel have taken advantage of the redirection of property taxes for schools and parks into the mayor-controlled Tax Increment Financing (TIF) slush fund to underwrite developers (Joravsky, 2015;Weber, 2003). Supposedly aimed at spurring development in "blighted" areas, almost half of the $1.3 billion in TIF funds allocated by Mayor Emanuel since 2011 have gone to development in downtown Chicago and gentrified surrounding areas (Joravsky & Dumke, 2015). Disinvestment and Disenfranchisement CPS's lack of investment in schools in Black and Latinx low-income communities contributes to overall disinvestment in the communities. Some schools lack even the basics of decent education, such as full time music and art teachers, up to date science labs, school nurses, adequate counselors, libraries, and textbooks for all students (Chicago Teachers Union, 2012;Lipman, et al, 2015). In December 2011, CPS's Chief Operating Officer stated that CPS did not intend to invest in schools that would be closed in the next 5-10 years (Ahmed-Ullah, 2011). In fact, schools serving low-income, working class African Americans, Latinx, and other students of color that CPS closed had previously endured cuts in resources and programs and multiple rounds of educational experiments and cycling of principals and district administrators, even in the face of community and parent interventions to support the schools (Gutierrez & Lipman, 2012;. A Dyett high School student protesting his school's phase out summarized: 8 The Center for Research on Education Outcomes (CREDO) 2015 study claimed "substantial" positive effects of attending a charter school. However, the study's methodology and conclusions have faced substantial criticism. Maul (2015) argues the actual effect sizes reported are very small, "well under a tenth of one percent of the variance in test scores" (p. 6), the methodology does not adequately control for differences between charter school families and those who do not choose charters, the "systematic exclusion" of lowerscoring students from the analyses, and ways the study estimates student growth. As proof of the continued disinvestment, it is now 2012, and now we only have one honors class…they put us on probation, they took away five teachers, and the worst part is now we are on the phase-out list…CPS always says it is in the interest of the children, but in reality, it is not (Lipman, Smith & Gutstein, 2012, p. 41). School closings and differentiated opportunities to learn are integral to a strategy that grooms targeted areas for devaluation and gentrification (Lipman, 2011)). In areas primed for redevelopment, closing schools facilitates displacement of the people who live there, making way to refurbish and rebrand them for a middle-class clientele. This is why, since CPS began closing schools en mass in 2004, parents and community members have charged that the survival of their communities is at stake. When CPS proposed the Mid South Plan in 2004, the Kenwood Oakland Community Organization charged it was "directly connected to gentrification of our communities" (Kenwood Oakland, n.d.). An African American parent we interviewed after the mass 2013 closings put it succinctly, "The reason they closing the schools so they can fade out our schools and open charter schools and push the people out of our communities" (Lipman et al, 2014, p. 23). Expansion of charter schools and outsourcing of school nurses and custodial and food services to private companies is part of the city's strategy to privatize public infrastructure and public labor. (Chicago also privatized parking meters, parking garages, the Chicago Skyway Bridge, and airport employment). Over the past decade, Chicago Public Schools lost roughly 30,000 students and added over 50,000 privately-run charter school seats (Black, 2013). The Board also turned over 31 schools, mostly on the African American South and West sides of the city, to the Academy for Urban School Leadership, a private operator contracted to manage CPS schools. In low-income neighborhoods, charter schools draw students away from public schools the state has neglected, hastening the closing of neighborhood schools and breaking the connections among parents, weakening the community. The loss of jobs for school support staff, who often live in the school neighborhood, also impacts the community economically CPS also instantiates neoliberal governance by appointed bodies and public private partnerships (Brenner & Theodore, 2002). The Board of Education is appointed by the mayor and makes decisions behind closed doors, with no accountability to those affected Lipman et al, 2014). In what has become an annual ritual, thousands of parents, teachers and students who pack gymnasiums and churches for public hearings are given two minutes to argue and plead for their schools-while those who make the decisions are not present. Candlelight vigils, town hall meetings, rallies, marches, sit-ins, sleep outs, school board takeovers, petitions, public testimonies, civil disobedience and other popular expressions of protest are the only recourse the public has to make its concerns heard. By putting schools on probation and expanding privately governed charter schools, CPS also eroded the power of Local School Councils, democratically elected local school governing bodies made up largely of parents and community members. In a school district that is 90% students of color, restriction of democracy signals that people of color are not capable of governing themselves. Outrage at this disenfranchisement is a primary engine of the campaign for an elected school board. Racism provides the material and ideological basis for these neoliberal strategies of what Harvey (2005) calls "accumulation by dispossession." On the one hand, colorblind market discourse obscures race as a factor in inequality, on the other, race is mobilized to demonize all that is public. The role of the state and private investors in the willful production of decayed public housing, disinvested neighborhoods, high unemployment, and "failing" schools is erased by a racialized narrative of concentrated poverty and Black social pathology (Imbroscio, 2008). Racially coded discourses of urban "cleansing" and "renaissance" of "bad neighborhoods" and "failing school" legitimate school closings and displacement of African Americans and some Latinx students (Lipman, 2009;Lipsitz, 2011). In turn, racialized disparagement of schools in Black and Latinx neighborhoods contributes to the construction of these communities as sites of violence and decay that must be remodeled for a new, more affluent, whiter, more "worthy" population (Lipman, 2009;Smith, 1996). These racially coded narratives mask the effects of racial segregation and racism and the disinvestment and churn neoliberal education, housing, and urban development policies have produced in low-income African American and Latinx neighborhoods. Austerity Urbanism CPS is also implicated in the austerity policies that Chicago adopted in the wake of the 2008 financial crisis. Like other local government officials, Chicago's mayor and the governor of the State of Illinois quickly reframed a crisis of the Wall Street banks and speculative investors as a crisis of public debt, declaring the necessity for fiscal restraint implemented through public austerity, cutting labor costs, and selling off public assets (Lipman, 2016;Peck, 2012). On its part, CPS opportunistically seized the crisis to accelerate school closings and privatization and to press for union concessions. CPS's risky investments in toxic interest rate swaps caught up with them as they lost millions of dollars to Wall Street banks. At the same time, years of failing to pay into the teachers' pension fund left the Board with millions of dollars of unfunded pension obligations (Gillers & Byrne, 2014;Martire, Otter & Kass, 2013;Roosevelt Institute, 2014). The Mayor's solution to the crisis created by these financial decisions has been to increase regressive sales and public service taxes, cut school budgets, outsource school services to private contractors, and increase teachers' share of their pension payments. Illinois' governor has refused to sign a state budget without cuts to public institutions and public sector union concessions to pay for the crisis. In short, the state is attempting to impose an "extreme economy" of "austerity urbanism" (Peck, 2012, p. 628) by socializing the debt to the banks. The CTU and progressive budget groups have intervened with a set of researched proposals to retrieve untapped revenue, including recouping money from the Wall Street banks for the toxic swap deals (Roosevelt Institute, 2014), drawing on the $1.36 billion in property taxes in the TIF fund (TIF Illumination Project), a small tax on financial transactions, and higher taxes on corporations and millionaires. Mayor Emanuel, the Illinois Governor, and the CPS Board have refused to consider these solutions. Austerity has hit schools across the board, but not equally. While some upper middle class white parents raised thousands of dollars to cover the loss of teachers and programs, schools in lowincome communities do not have the same advantage (Karp, 2015). They have had to make do without librarians, art and music teachers, and vital programs. In 2014, neighborhood high schools suffered the biggest budget losses with a third losing more than $1 million. Nearly every neighborhood high school on the African American South and far South Side had substantial cuts (Karp, 2014). A total of 80% of the schools designated to receive students from closed schools had budget cuts of more than $70,000-the average salary of one teacher (Karp, 2015). The Board's financial mismanagement and budget cuts have further eroded public confidence in CPS and increased support for an elected school board that is representative of Chicago residents. The CTU and community organizations have waged a sustained public campaign to challenge the narrative that there is no alternative to austerity. The Mayor and CPS' refusal to consider their common sense progressive revenue solutions and to cut school programs and staff instead, provides evidence for the CTU's claim that the mayor and the Board's interests lie with the banks and financial institutions rather than school children. Essentially a program of economic redistribution, the revenue campaign is an example of the maturing of the education movement from resisting to resistance and developing alternatives. Education Policy as Racist State Violence Neoliberal governments protect those who are valuable to capital and make vulnerable those who are not (Ong, 2006). African American and Latinx parents and students who testified at school closing hearings, and those who we interviewed charged the state with abandoning responsibility for their education (Lipman et al, 2014). A Walter Dyett High School student said, "It just feels like they don't care as much as we want them to. It's been heartbreaking. CPS just turned its back on us…." (Lee, 2013). The Board voted to phase out Dyett in 2012 after years of shrinking resources and programs and destabilizing the school's administration, despite parent and community organizing to bring resources into the school (Gutierrez & Lipman, 2012). Dyett was the last open-enrollment neighborhood high school in the African American Bronzeville area. When the Board voted to phase out Dyett, nearly a quarter of CPS' $423 million FY2015 capital budget was allocated to three selective enrollment high schools on the more white and affluent north side of the city (Lutton, 2014). Dyett exemplifies the state's abandonment of Black students' education, particularly in disinvested areas and areas being gentrified. School closings have turned some African American South and West side neighborhoods into what some parents call "school deserts." School Closings as State Violence Despite promises to provide additional support staff to help children deal with the chaos of the 2013 mass school closings, the CTU reported that CPS made no sustained investment in additional nurses, counselors, and social workers in the receiving schools (Caref, Hainds & Jankov, 2014). Teachers in receiving schools who CTU researchers interviewed reported their schools had received no special support for English Learners, students with Individual Education Plans, or homeless students. Our research on the closings (Lipman, Vaughan, & Gutierrez, 2014) also found that children suffered emotional trauma, classes were over-crowded, and some parents felt "locked out" of their children's new school. Parents disclosed that their children were "grieving" and they were grieving. These experiences help reframe the meaning of educational policies. It is critical to move beyond test scores and efficiency measures to grasp the human suffering, the pain and loss, that results from education policy experienced as racialized state violence (Dumas, 2014). A Bronzeville grandmother of 13 CPS students and a leading parent organizer describes school closings as "a hate crime." The collusion of the state and capital, the intertwined goals of capital accumulation and racial exclusion that dispossess people of color of their schools are constitutive of a larger spatialized process of organized state abandonment of low-income Black communities in particular (Lipsitz, 2014). Hulks of deteriorating empty school buildings scar Black and some Latinx neighborhoods and create safety hazards (Nitkin, 2015) in areas already facing high rates of home foreclosures, abandoned buildings, closed storefronts, and empty lots. Shuttering schools and forsaking buildings that were integral to community life and history is a declaration of the disposability of those who live there. When Price Elementary School in Bronzeville was closed in 2012, the Chicago Police Department took over part of the building for dog training. This was a daily insult to children who waited in front of their former school for a school bus that took them three miles to a school outside their neighborhood. The impact extends to the community as a whole. Schools are placemaking institutions, community anchors, centers of social networks and family resources and sites of collective memory, "homeplaces" in contexts of racial hostility (Haymes, 1995). Some closed schools were attended by generations of the same family. A parent whose child's school was closed said, "There are people who have lived in this community longer than I and watched kids come….So these are people that have long relationships with [closed school]. They family. Generations come out of this school" (interview with parent, quoted in Lipman et al, 2014, p.14). Prior to the massive waves of deindustrialization in the 1970s, the schools were institutional anchors of stable Black working class communities. At school closing hearings, while acknowledging historical racism, older community members recounted the days when "ducks swam in a pond" on the expansive green space of a South Side high school, and dedicated and talented Black teachers graduated young people who went on to be poets, musicians, and scientists. Even though the schools and communities experienced full-scale public and private disinvestment in the decades that followed, parents we interviewed frequently described their schools as "like families" (Lipman et al, 2014). Their involvement as unpaid teaching assistants, sports coaches, community liaisons, fundraisers, lunch room and playground supervisors, mentors, and crossing guards testifies to the meaning of schools in what Lipsitz (2011) terms the "Black spatial imagination," a commitment to public space as shared social space, a "public responsibility for which all must take stewardship" (p. 69). Schools that were closed were places of African American cultural and intellectual achievements and struggles for racial justice. As described above, Chicago was the site of massive school boycotts by Black parents in 1963 and 1964 against overcrowded, segregated schools and in 1968 Black students held walk-outs and sit-ins against racist curricula and teachers (Danns, 2014). They demanded community control of schools, Black history, more Black teachers and administrators, and general academic improvement of their schools. In the late 1980s, Black parents were again at the center of a grassroots movement that won elected Local School Councils in every CPS school. This was a measure of community control of local schools. Whatever has been achieved for Black children and teachers, whatever has been accomplished to make public schools into Black-led public spaces, has been wrested from the city's powerful interests through these collective struggles. Shuttering schools named for Mahalia Jackson, Benjamin Banneker, Marcus Garvey, Mary McCloud Bethune, Wendell Phillips, Walter Dyett, and more-against the will of the parents and students-does violence to that history. School closings instantiate the neoliberal state's deployment of violence to manage crises and contain and exclude "superfluous" populations (Camp, 2016). Chicago is also infamous for its deployment of police violence against Black, Latinx, and other marginalized people. 9 Since 2004, the city paid out $662 million in taxpayer dollars to settle police torture, misconduct, and wrongful conviction claims (How Chicago Racked Up, 2016). African American and some Latinx areas of the city are zones of police occupation and high rates of incarceration (Million Dollar Blocks, 2015) and 40% of the city budget is allocated to policing. In January 2017, a United States Department of Justice report (2017) castigated the Chicago Police Department for its pattern of excessive, unlawful, deadly force and put the Department under federal monitoring. Highly publicized police shootings of unarmed African Americans have provoked a new Black liberation movement led by young people. As I write this, the city's mayor and other public officials struggle to contain the political fallout from videotaped police murders of African Americans, police torture, and the findings of U.S. Department of Justice report. The anti-Black state violence of policing cannot be separated from the persistent violence of closing schools and abandoning the education of Black children. Dyett Hunger StrikeDisposability and the Struggle for Black Self-Determination Perhaps nothing more clearly demonstrated the connection between school policy, anti-Black state violence, and the spirit of self-determination than the Dyett hunger strike, a process I documented in my field notes and came to understand through conversations with community members and my direct participation. In August and September 2015, 12 African American parents, grandparents, teachers and supporters waged a 34-day hunger strike to reopen and transform Dyett High School. 10 In 2012, CPS began closing Dyett, the last open-enrollment neighborhood high school in Chicago's African American Bronzeville area. Bronzeville was the initial target of Renaissance 2010, the site of the demolition of thousands of units of public housing, and is experiencing extensive gentrification (Lipman & Haines, 2007). After two years of parents, students, teachers, and community members petitioning the Board and holding public meetings and protests, culminating in parents chaining themselves to a statue outside the Mayor's office, CPS promised to reopen the school as a neighborhood school and issued a request for proposals. Teachers, parents, university partners, and community organizations formed the Coalition to Revitalize Dyett which developed a comprehensive proposal for a community-driven neighborhood high school of Global Leadership and Green Technology (available from author). The school they envisioned was rooted in social justice and community involvement. It was a counter narrative to the dehumanizing, second class schools that Black children experienced in CPS. The proposal was the product of a vision developed by the community over two years through multiple community discussions and forums and was lauded by education experts, but at the last minute CPS violated its own process, making it clear that the community's plan would not be approved. After years of petitioning the Board and public officials, dozens of community meetings, and production of a 126 page proposal, Black parents, Dyett students, and community members were disregarded and dismissed. Feeling they had no further options, they took the extreme step of a hunger strike. The hunger strike garnered national attention and became both a demonstration of Black parents' determination to fight for quality public education-as they envision it-and a display of the state's apparent disregard for Black parents and children. Even as hunger strikers grew weaker and several were hospitalized, city officials did nothing, although doctors and nurses pleaded with the mayor to intervene in a life-threatening situation. At a press conference announcing the end of the hunger strike, a mother on hunger strike said, "It's been very disappointing that the mayor of Chicago will allow us to die even though we knew that he would do that because he's allowed Black children in our streets to die every day." Her statement linked CPS policies to the disposability of Black lives. When activists with the Movement for Black Lives marched from a rally against police murder to Dyett during the hunger strike, they connected police shootings of Black people with the anti-Black violence of Chicago's education policies. Dyett represents what the Journey 4 Justice Alliance calls "Death by a Thousand Cuts" (2014). The hunger strike won a partial victory. It forced CPS to reopen the school as an openenrollment neighborhood school, but not with the Coalition's framework. Yet parents and community members are still working to transform Dyett into the school they envisioned. The ongoing campaign is for a vision of education determined by the community itself, a claim to Black public space and to shape its future. The hunger strike illuminated education as an arena where the dialectic of racialized state violence and the struggle for Black self-determination and humanization plays out. "We are Fighting for the Soul of the City" 10 On day 24 of the strike they were joined by five more supporters and community members. Education is a key battleground for political power and the direction of the city as a whole. The policies of the mayor and CPS officials have alienated a broad section of parents and teachers. Cuts to school budgets, the Board's risky investments and ties to Wall Street Banks, and a series of corruption scandals 11 have undermined the credibility of the Board, CPS officials, and the Mayor. Inequitable distribution of resources, school closings, and privatization in Black and Latinx communities, exacerbated by the state's attempt to shift its budget deficit onto schools, have provoked more than a decade of grassroots organizing and city-wide opposition. Excessive high stakes testing has triggered an anti-testing backlash and student opt out movement, with much of the impetus coming from the very middle class parents CPS is trying to retain in the school system. The Board's imperious decisions, subject to no public oversight, have fueled broad public support for an end to mayoral control and an elected representative school board. This resistance is anchored by an alliance of two social formations-the Chicago Teachers Union and a multi-race, multi-class assemblage of community-based, educator, and parent organizations. This alliance, the Grassroots Education Movement (GEM), is the outgrowth of organizing by Black and Latinx community organizations against school closings beginning in 2004 and the CTU's transformation from a largely collaborationist business-style union into a social movement union allied with parents and students and other social movements (Weiner, 2012). One of the first things the new leadership of the CTU did after it was elected was to strategically allocate resources to organizing and research departments. And they established a "big bargaining team" of rank and file teachers to democratize and make transparent negotiations with CPS. Operating like a social movement, the union used the union newsletter to politically educate members, organized at the school level, and put the weight of its institutional resources behind mobilizing teachers to fight CPS policies in alliance with parents and students. Despite CTU's successful 2012 strike and the power and solidarity with parents and students it demonstrated, just eight months after the strike, protests involving thousands of parents and teachers were not strong enough to prevent the Board from closing 50 schools. And despite a mass city-wide one day walk-out in April 2016 that assembled an even broader labor-community coalition against budget cuts, the Mayor has not let up on his attacks on the CTU and attempts to socialize the debt by cutting school budgets and privatizing services and staffing and blaming teachers. In spring 2016, CPS privatized school nurses and expanded charter schools. Budget cuts continue. However, despite the many twists and turns of building a union-community coalition and its uphill battle against the powers of the mayor and entrenched political and corporate/banking establishment, CPS' racialized neoliberal narrative is no longer hegemonic. Accumulated crises and failed policies and persistent teacher-parent-student organizing, backed up by research, have precipitated a crisis of political legitimacy. This was demonstrated by two advisory referenda for an elected school board approved by almost 90% of voters (Chicago Board of Elections, 2012;Sabella & Cox, 2015). Education was at the center of the 2015 mayoral election. CTU President, Karen Lewis, had announced she was running and was ahead of incumbent Rahm Emanuel in the polls. Although she had to withdraw because of illness, the unpopularity of Emanuel's education policies was a key factor in his near-defeat for re-election, despite outspending his opponent, a former community organizer, 30-1. Polls in 2016 showed that three times as many Chicagoans trust the CTU than the mayor to improve education (Ruthhart & Perez, 2016). The state's legitimacy was further undermined by a string of notorious video-taped police shootings of African Americans and on-going police torture scandals. A militant Movement for Black Lives left the mayor and police department reeling. The CTU's progressive revenue solutions have gained traction, and the idea that taxing the rich is the only viable source of revenue has entered the public discourse (Patel & Kelleher, 2017). In essence, this is an opportunity for a politically progressive alternative to Chicago's neoliberal economic, racial, and social agenda, and the contest over public education is central. Looking back over the last decade, Chicago's grassroots organizing has become a social force to reckon with, despite the outsized power of Chicago's mayors in alignment with banking and corporate interests and despite losing many of the battles against school closings and the inevitable growing pains and complexities of an emerging union-community coalition. It has evolved from defensive actions to a proactive program for democratic governance (elected representative school board), revitalized and humanized public schools (CTU, 2012;Caref et al, 2015; the Dyett proposal), and economic redistribution to fully fund education. 12 Black and Latinx community organizations have provided much of the leadership and moral compass. In contract negotiation in fall 2016, the CTU won the Board's commitment to fund 20-55 sustainable community-centered neighborhood public schools, 13 to be developed by a taskforce consisting of CPS and GEM. Sustainable community schools are GEM's alternative to CPS' school closing/privatization agenda. As I write this in March 2017, negotiations are beginning over the character of these schools. This is a (contested) opportunity to create more equitable, supportive, and community driven schools that would be grounded in culturally relevant curriculum, meaningful assessment of student learning, community participation, wrap-around services, collaboration, and restorative justice. Another front in the battle to transform public education in Chicago. Concluding Thoughts The neoliberal political-economic-racial logics driving education in Chicago are national, indeed global, but they are shaped by local contexts. A set of local actors and institutions distinct to Chicago and its relations of social forces has shaped the dynamics of the contest over education policy in the city. Nonetheless, privatization and disinvestment in public schools and attempts to make teachers and public schools pay for the fallout of the financial crisis are playing out in Los Angeles, Detroit, Philadelphia, New York, New Orleans, Newark, and other cities-albeit in ways that are locally distinct. Grassroots organizations of parents and students and insurgent union movements in these and other cities are challenging neoliberal policy. A progressive caucus won the leadership of the Los Angeles teachers union, and parents in New York won sustainable community schools. The Journey 4 Justice, an alliance of mainly people-of-color grassroots community, youth, and parent-led organizations in 21 cities, has launched national campaigns to defend public education and put racial justice in the forefront. The CTU and community organizations have also amplified their power through national organizing with other teacher unions and parent and student organizations with similar goals. The experience in Chicago demonstrates interconnections between education policy and racial capitalism and its urban dynamics. The struggle over neoliberal education 12 By using Tax Increment Finance fund for schools, recouping the Board's losses from its toxic interest rate swap deals from the Wall Street banks, and a sales tax on financial transactions. 13 The model of Sustainable Community Schools was developed by the Alliance to Reclaim our Schools. Retrieved from http://www.reclaimourschools.org/sites/default/files/AROS%20Community%20Schools%20def_1.pdf policy is about state violence, claims to urban space, democratic participation, and political power. So what insights can we gain from Chicago's experience so far that might be useful in other places? First, neoliberalism is not a set of fixed policies, but a shape-shifting, opportunistic process of market-driven experimentation (Peck & Theodore, 2012). Neoliberalism is crisis-driven and crisis exploiting. The neoliberal state and corporate/financial actors opportunistically take advantage of crises generated by neoliberal policies to mutate and extend neoliberalization (Peck, 2012). For example, CPS' response to the fallout from the destabilizing effects of closing high schools was to create school turn-arounds, and the Board's response to the fiscal crisis it created was to expand privatization and essentially abandon schools in Black communities. Thus, the resistance has to make visible the larger racial neoliberal agenda behind specific policies. It has to educate the public, concretely, as the CTU and GEM are doing. Second, Chicago's experience makes clear that there cannot be education justice without racial justice. The assault on public education (Watkins, 2011) is a racial assault. This was made clear by the Dyett hunger strike. Chicago demonstrates that neoliberal education policy mobilizes white supremacy to further capital accumulation and re-inscribes racism, ideologically and materially. In Chicago, as elsewhere, Black and Latinx schools, teachers, and communities are bearing the brunt of disinvestment and state abandonment and are constructed as the problem to be fixed through markets and top-down accountability. In a context in which public schools have never been truly equitable and Black communities have never controlled their schools, racial justice is central to a counter-hegemonic education movement. The racialization of space and negation of Black humanity saturate the neoliberal urban education project, so it is not surprising that the resolve and social vision of Chicago's Black parents have often been the leading edge of resistance. This mirrors the experience in Detroit (Wilson, 2015) and the position of the Journey 4 Justice Alliance (see Journey 4 Justice, 2014). Third, Chicago's union-community alliance is the basis for whatever inroads have been made against CPS' agenda. The importance of social movement unionism cannot be overstated. The CTU's institutional capacity to organize, produce research, educate, and gain public voice for an anti-neoliberal anti-racist agenda have been pivotal in building an education movement in Chicago. The union's mass mobilizations and alliances with parents and students have shifted the terrain. The 2012 strike was a game changer. It demonstrated the power of teacher unions and made the CTU a force to reckon with. But the union also has much to learn from Black and Latinx parents and community organizations. In Chicago, Black and Latinx parents and community organizations have been the moral compass, the backbone, and often the political leadership of our movement. They have also been an essential support to the CTU. Nonetheless, there are inevitable race, class, and organizational disjunctions to be worked out between teachers and parents, unions and community organizations, people of color and whites. Building and strengthening community-education alliances on a foundation of shared values is complicated and unfinished. Finally, the outcome of many battles over education in Chicago has been uneven. But the measure of this contest cannot be calculated on a scorecard of "wins" and "losses." Many schools and veteran teachers have been lost, and children's education-especially Black and Latinx children-has been, and is being, damaged and their futures put at risk. Families and whole communities have suffered. The impact is irreparable and incalculable. There have also been victories (and partial victories): Dyett reopened as a neighborhood pubic school; CPS was pressured by parents and students to reduce standardized testing; sustainable community schools are being negotiated; some schools were saved from closing. All these gains were the product of endless hours of organizing and marching and sitting in and researching and sacrificing. Mothers and grandmothers have spent the night in jail. People have slept on the sidewalk in winter, and starved themselves, putting their long-term health at risk. Yet, as long as the present system and structures of power are in place, any gains are insecure. But what the education movement has accomplished, so far, in Chicago, is an accumulation of organizational strength and political consciousness and shared values and commitment. It has eroded the hegemony of neoliberal rationality and the legitimacy of the racial state and corporate/financial actors. Chicago's education movement has evolved from opposition to CPS policies to posing solutions to remake public education, including funding through economic redistribution, racial justice, democratization of governance, and a vision of an academically rich, culturally relevant, equitable holistic educational experience for all students. These are building blocks of a protracted counter-hegemonic struggle to transform the structures of power and ideologies governing public education. The Dyett hunger strike vividly demonstrated that the contest for the soul of public education is also a contest for, as Kevin Coval asserts, "A new city, a city anew, a city for all." Education is a strategic pillar of the neoliberal project to remake the city for capital accumulation and racial containment and exclusion. At the same time, the persistent organizing campaigns and counter narratives of an emergent grassroots movement have opened cracks in neoliberal hegemony. In the process, Chicago teachers and parents and students have inspired others who are facing a similar education landscape in other places. Researchers cataloguing how this dialectic plays out in various contexts may give us a fuller picture of an emerging, still largely localized, education movement and the social forces it is contesting. There are many critical questions: What are the shapeshifting dynamics of neoliberal policy making? How do education movements reframe racialized neoliberal discourses about urban public education? How do they sustain struggles against powerful economic and political actors when immediate gains are hard to come by? What can we learn about the possibility of building principled alliances of teachers, students, and parents that center racial justice and what alternatives do they develop to privatization and the historical inequities of public education? How is education related to political, economic, and social contests for the direction of the city? And what are the new challenges and possibilities of resistance in the regime of the 45 th president of the US? Pauline Lipman University of Illinois at Chicago plipman@uic.edu Pauline Lipman is Professor of Educational Policy Studies and Director of the Collaborative for Equity and Justice in Education at the University of Illinois at Chicago. Her multi-disciplinary research focuses on the political economy of urban education, particularly the inter-relationship of education policy, neoliberal urban restructuring, and the politics of race. She is the author of numerous journal articles and book chapters. Her most recent book is The New Political Economy of Urban Education: Neoliberalism, Race, and the Right to the City. Pauline is a core member of Teachers for Social Justice, which organizes with parents, teachers, students, and community organizations for education justice. She is co-author of the Data and Democracy Project and other community-collaborative research projects. About the Guest Editors Federico R. Waitoller, PhD. University of Illinois at Chicago fwaitoll@uic.edu Dr. Waitoller is an assistant professor in the department of special education at the University of Illinois at Chicago. His research focuses on urban inclusive education. In particular, he examines how neoliberal informed polices, such as top-down accountability, portfolio district models, and school choice converge with inclusive education efforts, and how these initiatives affect Black and Latino students with disabilities. His research also examines teacher learning efforts and pedagogies for inclusive education. Rhoda Rae Gutierrez University of Illinois at Chicago rrgutier@uic.edu Rhoda R. Gutierrez is a PhD candidate in Educational Policy Studies, Social Foundations and cofounder of the Collaborative for Equity and Justice in Education at UIC. Her research interests include neoliberal education policy and governance, impact of public school closings on families, teacher labor force restructuring, and activist research. She recently collaborated on a research project with parent and teacher activist-researchers to understand the impact of Chicago's historic 2013 school closings from the perspective of parents. Rhoda is a parent-activist with Parents 4 Teachers, which organizes for education justice in solidarity with teachers, students and community groups, and is involved in the citywide campaign for an elected representative school board in Chicago.	2019-05-11T13:03:06.655Z	2017-06-05T00:00:00.000	{ "year": 2017, "sha1": "025e75ee29a8b32e3a462af37ab9f216cb17f57c", "oa_license": "CCBYSA", "oa_url": "https://epaa.asu.edu/ojs/article/download/2660/1914", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "89302afec3cb42319af66cf80f23d434bb3d8bc4", "s2fieldsofstudy": [ "Education" ], "extfieldsofstudy": [ "Sociology" ] }
220522449	pes2o/s2orc	v3-fos-license	Effects of Rapid Recovery on Alcohol Hangover Severity: A Double-Blind, Placebo-Controlled, Randomized, Balanced Crossover Trial The aim of this study was to evaluate the efficacy of putative hangover treatment, Rapid Recovery, in mitigating alcohol hangover (AH) symptom severity. Using a double-blind, randomized, placebo-controlled, balanced crossover design, 20 participants attended the laboratory for two evenings of alcohol consumption, each followed by morning assessments of AH severity. Participants were administered Rapid Recovery and placebo on separate visits. In the first testing visit, participants self-administered alcoholic beverages of their choice, to a maximum of 1.3 g/kg alcohol. Drinking patterns were recorded and replicated in the second evening testing visit. In the morning visits, AH severity was assessed using questionnaires measuring AH symptom severity and sleep quality, computerized assessments of cognitive functioning as well as levels of blood biomarkers of liver function (gamma-glutamyl transferase (GGT)) and inflammation (high-sensitive C-reactive protein (hs-CRP)). There were no differences in the blood alcohol concentrations (BAC) obtained in the Rapid Recovery (mean = 0.096%) and placebo (mean = 0.097%) conditions. Participants reported significantly greater sleep problems in the Rapid Recovery compared to placebo condition, although this difference was no longer significant following Bonferroni’s correction. There were no other significant differences between Rapid Recovery and placebo. These data suggest that Rapid Recovery has no significant effect on alcohol hangover nor on associated biomarkers. Introduction Alcohol hangover (AH) is defined as the combination of negative mental and physical symptoms which can be experienced after a single episode of alcohol consumption, starting when blood alcohol concentration (BAC) approaches zero [1,2]. It is characterized by a general state of malaise and a range of physical and psychological symptoms including headache [3,4], fatigue [5], nausea [4] and reduced cognitive functioning [6][7][8]. These symptoms negatively impact daily activities such as driving [9,10], job performance [11,12] and studying [4]. AH is pervasive, affecting 75% of all social drinkers [13]. As well as subjective effects it contributes to significant economic costs. It is estimated that, due to associated absenteeism and presenteeism, AH costs the UK economy between £1.2 billion and £1.4 billion per year [14] (i.e., approximately US $1.5 to $1.7 billion) and the Australian economy over AUS $3 billion annually [15] (i.e., approximately investigated the effects of Korean pear in a sample of 14 healthy male-only Asian subjects. Therefore, more research is needed to confirm these findings in groups of non-Asian descent men and women. The aim of this investigation was to examine the effects of Rapid Recovery on AH symptom severity, inflammation, sleep quality and cognitive functioning. Rapid Recovery is an oral capsule that contains the amino acid L-cysteine and B and C group vitamins. It is proposed by the manufacturers that these ingredients will improve acetaldehyde metabolism and reduce oxidative stress. L-cysteine plays a role in reversing oxidization in the liver, with animal research showing that L-cysteine accelerates the breakdown and reduces the accumulation of acetaldehyde [51]. Another rodent study found that the administration of L-cysteine combined with vitamins B-1 and C reduced mortality caused by acetaldehyde poisoning [52]. In the current study, we tested the hypothesis that Rapid Recovery would reduce AH severity in social drinkers. A number of relevant biomarkers were co-monitored. Method This study was conducted in accordance with the Declaration of Helsinki and was approved by the Swinburne University Human Research Ethics Committee (SUHREC, 2018/275). This study was registered with the Australian New Zealand Clinical Trials Registry (ANZCTR, ACTRN12618001996257). Design This study was a semi-naturalistic, randomized, double-blind, placebo-controlled, crossover clinical trial. The laboratory was set-up to simulate a bar-like environment and participants consumed alcoholic drinks of their choice and at their own pace, to a maximum of 1.3 g/kg alcohol. Participants were administered either placebo or active treatment over two testing visits. Participants Twenty-three participants who were healthy, aged 21-50 years old and regularly experienced hangovers were enrolled in the study. Three participants withdrew at the first morning visit, two withdrew due to illness and one failed to meet the eligibility requirement of a (BAC of 0.00% at the morning visit. The final sample consisted of 20 participants (65% female) with a mean age of 30.30 years (range 25-43 years old). All participants were free of any current or history of drug or alcohol abuse, medically treated liver or renal impairment, pregnancy or breast feeding in females, and current use of any medication that could potentially affect the outcome of the study. Breath Alcohol Concentration (BAC) BAC was measured at the beginning of each testing visit to ensure a reading of 0.00%. In the evening testing visits, BAC was measured approximately 20-min after the final alcoholic drink. BAC was collected using a regularly calibrated Lion Alcolmeter SD400PA. Assessment of Hangover Severity Overall hangover severity was measured using a single one-item rating and severity of 23 hangover symptoms were rated on an 11-point Likert scale ranging from 0 to 10, with higher scores indicating a more severe hangover. The 23 items were derived from the Alcohol Hangover Severity Scale, the Hangover Symptoms Scale and the Acute Hangover Scale [53,54]. This composite scale has been successfully implemented in previous hangover research [55]. Sleep Quality Assessments Self-reported assessments of sleep quality comprised of the Groningen Sleep Quality Scale (GSQS) [56] and the Karolinska Sleepiness Scale (KSS) [57], which measured sleep quality during the previous night and current sleepiness, respectively. The GSQS comprises of 15 sleep complaints requiring a "yes" or "no" response indicating whether they had been experienced during the previous night's sleep. Scores range from 0 to 14, (the first item is not scored) with higher scores indicating poorer sleep quality. The KSS requires participants to indicate their level of fatigue in the last five minutes on a single-item using a nine-point Likert scale. Higher scores indicate greater levels of sleepiness. These scales have been implemented successfully in previous hangover research [9,55]. Assessment of Biomarkers for Inflammation and Liver Function High-sensitivity C-reactive protein (hs-CRP) tests were used to measure inflammation and gamma-glutamyl transferase (GGT) tests were used to measure liver function. Assessment of Cognitive Performance Cognitive performance was measured using the following tests available on the Vienna Test System (Schuhfried GmbH, Moedling, Austria). This test system assesses cognitive functioning that influence driving ability. The entire battery required approximately 15-20 min to complete. Reaction Test (RT) This test measures reaction time and motor time in response to optical and acoustic signals [58]. Participants were asked to place and leave their index finger on a pressure-sensitive key (i.e., rest key). Using the same index finger, participants were required to react as quickly as possible to the signals by pressing a target key before retiring their finger to the rest key. Performance was measured according to mean reaction time, mean motor time and number of correct reactions. Determination Test (DT) This test assesses reactive stress tolerance, divided attention and mental flexibility [59]. Participants are presented with various visual and auditory stimuli and are required to respond the stimuli by pressing corresponding response buttons with either their hands or feet, using the response panel and foot pedals of the Vienna Test System. Performance was assessed according to reaction time, number of correct responses, number of errors and number of missed responses. Adaptive Tachistoscopic Traffic Perception Test (ATAVT) This test assesses visual observation skills, visual orientation ability, speed of perception and skills in obtaining a traffic overview [60]. Images of traffic situations appeared briefly on a computer screen and the participant was asked to state what was in each image, by choosing from five answer options; motor vehicle, road sign, traffic light, pedestrian and bicycle. Performance was measured according to reaction time and the number of errors made. Perceived Treatment Order Awareness of the allocated condition order (active-placebo or placebo-active) was measured at the end of the trial. Participants were asked which treatment (active or placebo) they believed they had received on the first and second testing visit. Procedure Prior to undergoing any testing procedures, participants provided written informed consent and were assessed for eligibility. Participants then underwent training and practice in completing the RT, DT and ATAVT tasks, and provided a baseline blood sample for hs-CRP and GGT analyses. All testing visits were held in the laboratory, with intoxication visits held between 17:00 and 00:30, and hangover visits held the following morning between 7:00 and 11:00. The two evening visits were held within 7-14 days of one another. During the evening visits, the laboratory was set-up to mimic a bar and background music was played while participants socialized with one another. Participants were advised to avoid alcohol for 24-h prior to the intoxication visits, food and drink (other than water) for 2-h prior to all testing visits and alcohol, drugs, food and caffeine between the evening and morning visits. At the beginning of each evening visit, participants were provided with a meal, the type and quantity of food consumed in the first visit was recorded and replicated in the second evening visit. Participants were then instructed to freely consume the drink type(s) of their choice (of wine, cider, beer, spirits), to a maximum of 1.3 g/kg alcohol. The time that each drink was started and finished was recorded and drinking behavior was replicated in the second evening visit. Participants were administered the first dose of the study treatment with their final drink and were provided the second dose to self-administer upon their first awakening the following morning. The study treatment was either placebo (corn flour) or Rapid Recovery (L-cysteine, thiamine, pyridoxine and ascorbic acid). The contents of the study treatments were controlled by a laboratory independent of the manufacturer. Participants returned to the laboratory the following morning where they were initially breathalyzed to ensure a BAC reading of 0.00%. Once deemed eligible, participants were able to commence the testing procedures. Statistics and Analyses Statistical analyses were conducted using SPSS, Version 25 (IBM Corp, Armonk, NY, USA). All variables were analyzed using paired sample t-tests comparing Rapid Recovery with placebo. The sleep quality assessments and Hs-CRP and GGT levels were correlated with overall hangover severity. Lastly, a chi-square test was used to determine whether there was a significant difference between correct and incorrect perceived treatment order. In order to further investigate whether the obtained data was more in favor of the null hypothesis (H0, i.e., the assumption of no differences between the active and placebo condition) or more in favor of the alternative hypothesis (H1, i.e., the assumption of differences between the active and placebo condition), add-on Bayesian statistics were conducted using the standard settings of SPSS for the respective tests. Based on the cutoffs suggested by Wagenmakers, et al. [61] the Bayes factor (BF) of 1 does not provide evidence for either hypothesis. Larger BF values provide stronger evidence for the H0 (compared to the H1), while smaller BF values provide stronger evidence for the H1 (compared to the H0), given the obtained data. Specifically, values 1-3 (1/3-1) are seen as anecdotal evidence for the H0, values 3-10 (1/10-1/3) are seen as substantial evidence for the H0, values of 10-30 (1/30-1/10) are seen as strong evidence for the H0, values of 30-100 (1/100-1/30) are seen as very strong evidence for the H0, and values of >100 (<1/100) are seen as extreme evidence for the H0. Hangover Symptom Severity The only hangover symptom to significantly differ according to testing condition was "sleep problems" t(19) = 2.10, p = 0.049, with more severe sleep problems in the active (mean = 2.59, sd = 2.86) compared to placebo (mean = 1.63, sd = 1.75) condition. Following Bonferroni's correction, this difference was no longer significant. Add-on Bayesian analyses for the non-significant effects revealed that in most cases, the obtained BF provided substantial evidence for the H0, as indicated by BF values between 3 and 10. For the other factors ("reduced appetite", "sweating", "heart beating" and "vomiting"), the Bayesian analyses still provided anecdotal evidence in favor of the H0, as indicated by BF values between 1 and 3. Taken together, all of these findings support the assumption that none of the investigated measures improved during the active condition. Hangover symptom severity scores can be found in Table 1, below. Table 1. Hangover symptom severity scores (means and standard deviations) in the Rapid Recovery and placebo conditions. Descriptive data is given in the left columns, while the p value obtained from paired samples t-tests and the Bayes factor (BF) value obtained in case of non-significant differences (i.e., p values < 0.05) are provided in the right columns. Sleep Quality and Cognitive Performance There were no significant differences between the Rapid Recovery and placebo conditions on the Groningen Sleep Quality Scale (GSQ), Karolinska Sleepiness Scale (KSS), reaction test (RT), determination test (DT) and adaptive tachistoscopic traffic perception test (ATAVT). The mean scores and standard deviations are displayed in Table 2, below. Self-rated overall hangover severity significantly correlated with GSQ (r = 0.552, p = 0.012) and KSS (r = 0.764, p ≤ 0.001) scores in the placebo condition. The same result was found in the treatment condition, with overall hangover severity scores significantly correlating with GSQ (r = 0.638, p = 0.002) and KSS (r = 0.762, p < 0.001) scores. Table 2. Sleep quality and cognitive performance scores (means and standard deviations) in the Rapid Recovery and placebo conditions. Descriptive data is given in the left columns, including Groningen Sleep Quality Scale (GSQS), Karolinska Sleepiness Scale (KSS), reaction test (RT), determination test (DT) and adaptive tachistoscopic traffic perception test (ATAVT) while the p value obtained from paired samples t-tests and the BF value obtained in case of non-significant differences (i.e., p values < 0.05) are provided in the right columns. Item Rapid Levels of Biomarkers for Inflammation and Liver Function There were no significant differences in hs-CRP and GGT levels in the Rapid Recovery compared to placebo condition (all p ≥ 0.376). Bayesian add-on analyses further provided substantial evidence for the H0 (all BF ≥ 3.587), thus demonstrating that both measures did not differ across conditions. Furthermore, hs-CRP and GGT levels did not significantly correlate with self-rated overall hangover severity in either of the testing conditions (all p ≥ 0.286). For both conditions, add-on Bayesian analyses provided substantial evidence for the lack of correlation between hs-CRP and overall hangover ratings (all BF ≥ 3.000) and for the lack of correlation between GGT and overall hangover (all BF ≥ 3.008). The hs-CRP and GGT levels can be found in Table 3, below. Table 3. High sensitivity C-reactive protein (hs-CRP) and gamma-glutamyl transpeptidase (GGT) levels (means and standard deviations) at baseline and in the Rapid Recovery and placebo conditions. Descriptive data is given in the left columns, while the p value obtained from paired samples t-tests comparing the placebo and active condition, as well as the BF value obtained in case of non-significant differences (i.e., p values < 0.05) are provided in the right columns (N = 16). Percieved Treatment Order and Adverse Events A total of 60% of the participants guessed the correct condition order, indicating adequate blinding, χ 2 = 0.80, p = 0.371. There were no reported adverse events associated with Rapid Recovery. Discussion The current study assessed the effects of Rapid Recovery on hangover symptom severity. The hypothesis that Rapid Recovery would reduce AH severity was not supported. There were no significant differences between placebo and Rapid Recovery on self-rated overall hangover severity, sleep quality, CRP and GGT levels and cognitive performance. Furthermore, Bayesian add-on analyses provided credible evidence that the assumption of a null effect was more likely (than the assumption of non-significant/residual differences), given the obtained data. Of the 23 hangover symptoms that were assessed, the only significant difference between Rapid Recovery and placebo was found on the symptom of "sleep problems", which was worse following Rapid Recovery administration. The ineffectiveness of Rapid Recovery to reduce AH severity may indicate that administration of l-cysteine combined with B and C vitamins does not improve acetaldehyde metabolism, or that acetaldehyde is not responsible for AH severity. The results of this study further support the relationship between poor sleep quality and hangover severity [8,[62][63][64], with significant and positive correlations between AH severity, and poor previous night's sleep quality and current sleepiness. The current study failed to provide any evidence for a correlation between AH severity and CRP or GGT levels, which remained within normal ranges at each testing timepoint. Currently, the evidence for an association between AH severity and CRP is mixed, with some studies [25,39] demonstrating support for an association, while one other study [40], consistent with the findings of the current study, failed to find significant correlation between AH severity and CRP. While GGT is a reliable biomarker of liver damage caused by chronic heavy drinking [65][66][67], previous research has indicated that GGT levels are not associated with AH susceptibility [49] and, consistent with the findings of this study, are not necessarily elevated during AH [39]. Although it was not demonstrated in the current study, compelling evidence indicates an impairing effect of AH on immune functioning, and more reliable markers of this may include, but are not limited to, interleukin (IL)-6, IL-10, IL-12 and tumor necrosis factor (TNF)-α [37,68,69]. This investigation utilized a novel, controlled and ecologically valid methodology, which was found to successfully induce AH. Participants obtained mean BACs of 0.096% and 0.097% in the Rapid Recovery and placebo conditions, respectively, levels beyond that required to induce a hangover [70]. By enabling participants to self-administer alcohol within a controlled laboratory setting, we were able to overcome several commonly occurring methodological issues within the area of AH research. While methods of alcohol dosing used in previous laboratory studies assessing AH have been criticized for not mimicking real-life drinking behaviors, naturalistic studies have been criticized for lacking experimental control, and relying on self-reported alcohol intake to calculate estimated BAC [4,[71][72][73]. The current study used methodology which combined the advantages of naturalistic approaches (i.e., participants drinking alcohol of their choice in a social setting) with those of laboratory studies (i.e., a controlled environment, objective measures of BAC and other biomarkers, veracity of treatment administration). There were several limitations in this study. Firstly, we allowed participants to consume their preferred type of alcohol to ensure drinking behaviors replicated real-life drinking. Although drinking behaviors were consistent across the two testing visits, which eliminated intraindividual differences, this introduced interindividual variability. Alcoholic drinks contain various concentrations of congeners, which have been found to increase AH severity [17,74]. On the other hand, the fact that each individual's session was matched, somewhat, mitigates against this influencing our results. Furthermore, it is possible that the effectiveness of Rapid Recovery is dependent on individual factors, for example, genetic variations or tolerance to alcohol. This was evident in literature on the effectiveness of Korean pear juice in treating certain AH symptoms in particular genetic subgroups but not others [46]. The sample size of this study was too small to allow meaningful subgroup analysis. Lastly, we did not assess hangover symptoms following alcohol abstinence because, although interesting, this was not necessary for the aim of this study, i.e., comparing Rapid Recovery and placebo. As such, we are unable to determine the severity of AH obtained in this study. While mean BAC levels are beyond those deemed required to induce a hangover [70], the mean overall hangover severity score is relatively low. Recent evidence indicates that BAC may not be the most appropriate predictor of AH severity, which is better predicted by levels of subjective intoxication and increased alcohol consumption compared to usual [70]. While these factors were not assessed in the current study, future AH research should aim to include measures of subjective intoxication and typical alcohol intake. In conclusion, the findings from this study suggest that the administration of Rapid Recovery does not mitigate AH severity, and Hs-CRP and GGT levels are not associated with AH. Further research is required to assess the impact of an effective hangover treatment on alcohol consumption and determine whether it would encourage excessive drinking. Importantly, an effective hangover treatment would not mitigate all adverse factors associated with heavy drinking, such as chronic disease and injury. The development of an effective hangover treatment is currently hindered by a lack of understanding of the pathology of AH. As such, future research should continue to assess the pathology of AH to enable the development of treatments that target key mechanisms involved in the AH. Conflicts of Interest: Over the past 36 months, A.S. has held research grants from Abbott Nutrition, Arla Foods, Bayer, BioRevive, DuPont, Fonterra, Kemin Foods, Nestlé, Nutricia-Danone, and Verdure Sciences. He has acted as a consultant/expert advisor to Bayer, Danone, Naturex, Nestlé, Pfizer, Sanofi, Sen-Jam Pharmaceutical, and has received travel/hospitality/speaker fees from Bayer, Sanofi, and Verdure Sciences. Over the past 36 months, J.C.V. has held grants from Janssen, Nutricia, and Sequential, and acted as a consultant/expert advisor to Clinilabs, More Labs, Red Bull, Sen-Jam Pharmaceutical, Toast!, and ZBiotics. S.B. has received funding from Red Bull GmbH, Kemin Foods, Sanofi Aventis, Phoenix Pharmaceutical, BioRevive, Australian Government Innovations Scheme and GlaxoSmithKline. A.K.S. has received funding from Daimler and Benz.A.J.A.E.V.D.L. has no conflicts of interest to declare. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.	2020-07-15T13:05:57.914Z	2020-07-01T00:00:00.000	{ "year": 2020, "sha1": "d0f96e42dc0b6fe77bc9a30662b812098f27b6d4", "oa_license": "CCBY", "oa_url": "https://res.mdpi.com/d_attachment/jcm/jcm-09-02175/article_deploy/jcm-09-02175-v2.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "2c6042ffef85755bfed71ce73d2a4027a83958fe", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
263609201	pes2o/s2orc	v3-fos-license	Implementation of hyperspectral inversion algorithms on FPGA: Hardware comparison using High Level Synthesis Hyperspectral imaging is gathering significant attention due to its potential in various domains such as geology, agriculture, ecology, and surveillance. However, the associated processing algorithms, which are essential for enhancing output quality and extracting relevant information, are often computationally intensive and have to deal with substantial data volumes. Our focus lies on reconfigurable hardware, particularly recent FPGAs. While FPGA design can be complex, High Level Synthesis (HLS) workflows have emerged as a solution, abstracting low-level design intricacies and enhancing productivity. Despite successful prior efforts using HLS for hyperspectral imaging acceleration, we lack a comprehensive research to benchmark various algorithms and architectures within a unified framework. This study aims to quantitatively evaluate performance across different inversion algorithms and design architectures, providing insights for optimal trade-offs for specific applications. We apply this analysis to the case study of spectrum reconstruction processed from interferometric acquisitions taken by Fourier transform spectrometers. Introduction Hyperspectral imaging has gained considerable interest both in the industry and the scientific community for its promising capabilities in a wide range of fields, such as geology, agriculture, ecology and surveillance [2,35,25].However, the associated processing algorithms needed to improve the quality of the user-end product or to extract information of interest from the captured data are generally computation-intensive [23] and run on large data volumes [22].These issues are becoming even more relevant in more recent applications, as the dimensionality of hyperspectral data has increased with finer temporal resolutions made available by novel technologies [23].Therefore, fast and efficient computation is required to perform the necessary processing [32,40,16,30,7], especially for onboard applications.In this study, we focus our attention over reconfigurable hardware; these devices have became a preferred target for digital signal processing (DSP) applications, as they can be more easily mass-produced in comparison to equivalent efficient application specific integrated circuit (ASIC) implementations.In particular, recent field-programmable gate arrays (FPGAs) show increased hardware capabilities at an incrementally decreased cost, making them competitive for many applications that were originally implemented over ASICs [38]. However, using FPGAs architectures with low level design flows, such as manually coded register transfer levels (RTLs), significantly decreases the designer's productivity, as they prevent an exhaustive exploration of the design space.Moreover, RTLs are usually technology-dependent, which makes integrated processors (IPs) difficulty portable [8].All these reasons have made it necessary to adopt high-level synthesis (HLS) design workflows that keep up with the design complexity [3].Abstracting the architecture of RTL can in fact vastly speed-up the design process, allowing to explore multiple design choices (i.e.parallelism, frequencies) to make reasonable trade-offs.By choosing the appropriate optimizations in the most advanced HLS tools, hardware designers can generate circuits with a precise control of the performance-resource trade-off while ensuring a high productivity [27].HLS tools have the advantage to allow the designer to focus on the algorithms and automatically infer the low-level hardware architecture, while also giving him the possibility to drive low-level features when mandatory. Despite the promising results demonstrated by previous studies in using HLS for accelerating hyperspectral imaging algorithms [1], there is still a lack of a comprehensive research that benchmarks different algorithms and architectures using a unified framework.Domingo et al. [5] evaluated the use of Intel OpenCL SDK for accelerating spectral classification in hyperspectral imaging using K-Nearest Neighbour as a test case with several optimization techniques, including loop unrolling, kernel vectorization, and bitwidth optimization.Lei et al [22] proposed a deep pipelined structure for a subpixel target detection algorithm, called constrained energy minimization (CEM), and improved the efficiency of the algorithm by breaking the computation into multiple stages to achieve higher pipeline acceleration.[10] proposed an RTL-based implementation of the hyperspectral N-FINDR, a data analysis algorithm able to extract subpixel information on the materials present in the scene.The authors optimized the performance by including a direct memory access (DMA) module and using a pre-fetching technique to hide the input/output interfacing latency. In this work, we aim to provide a quantitative evaluation of the performance of different inversion algorithms and design architectures, so the most suitable trade-off can be identified for a given application and design a hardware prototype that meets the performance and cost requirements. The selection of a proper inversion algorithm is particularly crucial for hyperspectral image reconstruction, where the development of custom hardware solutions can be time-consuming and costly.We tackle these problems in this work by analyzing a specific case study of spectrum reconstruction for interferometric acquisitions taken by Fourier transform spectrometers (FTSs) [11].For such devices, for which the Michelson interferometer is the most well-known example, the acquisition, known as interferogram, is captured in the Fourier domain instead of the original one.A straightforward strategy to recover the information in the original domain consists in taking the discrete cosine transform (DCT) of the acquisition.While efficiently implemented on hardware [33] and through fast Fourier transform (FFT) [24], various limitations apply in practical scenarios.For example, the FFT requires the interferogram to be regularly sampled, and most importantly, the Fourier transform often does not accurately describe the transfer function of many FTSs, such as in the case of the Fabry-Perot interferometer.For such reasons, various alternative solutions have been proposed in the literature to deal more efficiently with such reconstructions.Some widespread examples, such as those based on penalized singular value decomposition (SVD) [14,37], do not have equally commonplace hardware implementations in FPGAs or DSP. The main contribution of this paper is to provide a hardware-centric implementation benchmark of various inversion methods for hyperspectral image reconstruction.This benchmark is motivated by the difficulty to explore large design space and to be able to choose the appropriate method that fits the requirements as well as the constraints.The novel content of our proposed benchmark can be summarized as follows: -We propose an improvement of the Coley-Tukey implementation of the FFT with a pre-and post-processing normalization step for each computational block.This stage allows to share the exponent of the block floating point (BFP) between each FFT stage, resulting in accelerated calculation of about 3 times.-We conduct a hardware comparison of different hyperspectral reconstruction algorithms by evaluating their resource overhead and acceleration gain.This comparison can help establish a new criterion for choosing between these methods, with easily interpretable benefits and limitations to their extension to real-world use cases.-We investigate the impact of arithmetic approximations on the quality of the hyperspectral reconstruction.We analyze the fixed-point data width and its effects on both reconstruction quality and hardware resource utilization.By highlighting the impact of arithmetic approximations on reconstruction quality, this research can help identifying the optimal trade-off between quality and efficiency for a given hyperspectral imaging application. The remainder of the paper is organized as follows: in Section 2 we describe the case study and some of the algorithmic implementations used to solve it, in Section 3 we discuss their hardware implementations and the relevant criteria to compare their performances, and in Section 4 we describe the related experiments. Spectrum reconstruction techniques In this section we address the problem of spectrum reconstruction from the acquisition of FTSs by describing some classes of available approaches from the literature.In particular, Section 2.1 describes the optical theory behind the formation of an interferogram, Section 2.2 describes the procedure of spectrum recovery through Fourier inspired techniques, while Section 2.3 describes those based on the SVD. Problem statement The FTS defines a family of optical devices which measures a spectrum by exploiting the principle of interferometry, that is, by combining a set of coherent waves which travel different paths, whose difference, in term of optical length, is known as the optical path difference (OPD) δ.The raw acquisition of the device, known as interferogram, is obtained by varying the OPD and requires a step of digital processing to retrieve the spectrum in its original domain. In order to capture an interferogram, the FTS implements a mechanism to split the incoming light wave into multiple interfering components associated to different traveled paths.Different approaches to achieve this goal allow to distinguish among different classes of FTSs [11].The Michelson interferometer shown in Figure 1 is the most common example of FTS.The device uses a half silvered mirror to split the incoming wave into a directly transmitted one and one that is reflected over a second mirror surface.The optical path difference between the split waves can be adjusted by modifying the relative distance or angle of one of he reflective mirrors.Assuming no loss of coherency, only a phase shift is introduced among the split waves, and the interference is measured when they recombine over a focal plane. In mathematical terms, the k-th acquisitions y k associated to the k-th OPDs δ k can be modeled as [34]: -S(σ) denotes the spectrum to reconstruct.For simplicity, we express the spectrum in terms of wavenumbers σ, the reciprocal of the wavelength; -B is the bandwidth of the instrument in the wavenumber domain; -T (σ, δ k ) is the transmittance response of the device for a given OPD δ k .This transmittance includes various device-specific mechanisms such as reflections, refractions, attenuations, integration time, and so on. The vector y = {y k } k=[0,...,M −1] can be interpreted as a sampled version of an ideally continuous interferogram, and the goal of the reconstruction is to obtain an estimation x of the digital representation x = {x n } n∈[0,..,N −1] of S(σ).This representation can be obtained as a numerical integration of the spectrum over N intervals of length ∆σ = B/N , that is: where, assuming that T can be approximated by its value in the midpoint of each interval, yields: The model described above is relevant for singlepixel acquisitions; various designs have been proposed for imaging systems based on this principle, where the sensor on the focal plane is capable to capture the interferogram associated to different points on the targeted scene.A common example for such cases involves devices based on Fabry-Perot interferometry, as samples of the interferogram is obtained by allowing the light rays to follow different paths within an etalon with parallel faces.Such designs may for example involve single-interferometers with diffraction gratings [21], or more commonly multi-aperture snapshot spectrometers where interferometers with different thicknesses are disposed over a staircase matrix [28], and the raw acquisition is composed of several subimages.This image formation model is described in detail in [31], with several algorithmic solutions that have been proposed for optimizing the raw acquisitions [18], a subset of which is addressed in this work. In the following sections, we will focus only on the situation where the acquisition y represents a singlepixel interferogram; for multi-aperture devices, this is obtained by selecting matching pixels across different subimages.The estimation of the spectrum x is related to a single point on the scene and obtained by applying the chosen inversion algorithm over a vector y. FFT-based inversion The family of strategies based on the transformation in the Fourier domain assumes that the transfer function T (σ, δ k ) of eq. ( 1) is given by a specific analytical expression. To derive such expression, let us assume that incident spectrum S(σ) is split into two interfering components S 0 and S 1 , whose optical intensities are respectively S 0 (σ) = aS(σ) and S 1 (σ) = arS(σ)e jϕ .Here, the coefficients a and r denote the subsequent attenuation factors of each of the two waves, while ϕ = 2πσδ k is a phase shift introduced by the FTS, which depends on its operating principle. The interferogram y k is then given by integrating the collected energy over the whole wavenumber range: which can be rewritten as: where S = B 0 S(σ) dσ is the average value of input spectrum and S(\|σ\|) is its even symmetrical extension of S(σ) to negative wavenumbers.If we assume that S is either known (i.e., measuring it for an OPD equal to 0) or zero (i.e., if the spectrum is low pass filtered before the measurement), and a and r are known, the inverse Fourier transform of y k allows to recover the desired spectrum.By performing the numerical integration (3), eq. ( 5) becomes: where the second step assumes that the OPDs is regularly sampled with a step size ∆δ, or in other words δ k = k∆δ.Eq. ( 6b) is equal to the DCT-II expression up to a scaling factor, assuming ∆σ∆δ = 1 and N = M .Hence, the spectrum reconstruction can be performed with an inverse DCT.The DCT can be implemented efficiently with slight variations on the FFT transformation (i.e. through Makhoul formula [24]).However, the interferogram is rarely sampled regularly in practice, which typically demands the interferogram to be interpolated over equally spaced OPDs as pre-processing step. SVD-based methods In this family of techniques, the model of eq. ( 3) is defined as a linear transformation: In the above equation, the transfer matrix A ∈ R M ×N is generally not square and whose coefficients a kl are given by: Compared to the previous case, this formulation allows for more generality, as it allows to express any type of linearizable transfer function, e.g. in the case of Fabry-Perot interferometers, this is given by the Airy distribution [15]: where a is an attenuation factor and r is the reflectivity of the Fabry-Perot etalon, while δ k is the OPD introduced by a single round trip reflection within the cavity. In the approach of linear regression, the estimation x of x is set up as the solution of the following problem: where ∥•∥ 2 denotes the ℓ 2 norm and The pseudoinversion is efficiently implemented through the SVD on the matrix A and then taking the reciprocal of the obtained singular values (s.v.s) [36].In other terms, this is a two steps operation: where U and V are semi-orthogonal matrices and Ξ is a diagonal matrix, whose elements on the main diagonal {ξ r } r∈[1,...,R] , are ordered in increasing order.The matrix Ξ −1 is still diagonal and the elements on the main diagonal {1/ξ r } r∈[1,...,R] are the reciprocal of those of Ξ. Despite its simple mathematical formulation, the closed form solution of eq. ( 10) holds almost no practical value in real scenarios , as the problem is ill-posed or ill-conditioned in the sense of Hadamard.Specifically, the multiplication by A † enhances the noise included in the acquisition, causing instability in the results.To avoid this issue, a widespread approach is to employ a modified version Ξ ′ of Ξ −1 [39], whose singular values are penalized.The estimation is hence obtained as: The s.v.s {ζ r } r=[1,...,R] of Ξ ′ can be chosen with techniques such as the truncated singular value decomposition (TSVD) [14] or the Tikhonov regularization (TIK) [37], defined as follows: In the above equations, the matrix rank 1 ≤ R ′ ≤ R and the ridge regression parameter λ ≥ 0 are userdefined, and the pseudo-inverse (PINV) is obtained as a special case when R ′ = R and λ = 0, respectively.The parameters R ′ and λ are usually chosen according to the amount of noise present in the system [9]; small values of R ′ or large values of λ introduce a stronger penalization which allows to compensate to smaller signalto-noise ratio (SNR) scenarios.For optical device operating in different conditions (e.g., lighting, temperature, environmental conditions), these parameters have to be set dynamically.For example, they can be selected from a set of possible choices, or be estimated from the characteristics of the acquisition.Techniques such as the L-curve method [13] have tried to address this problem, but with strong limitations [12].A hardware implementation which is able to provide a quick inversion regardless of the choice of such parameter is an important asset at the consumer's disposal to improve the quality of the reconstructed signals; we aim to address this issue in this paper. Hardware implementations This section covers the hardware implementation of the described algorithms.For hardware implementation, the most limiting factor for accurate computation involve computation noise.This noise refers to errors in computation due to hardware imperfections or algorithmic limitations, for which numerical stability is a desirable characteristic of our implementation to assure that those errors do not lead to significant inaccuracies on the final results.As a priority, we want to ensure that the implemented algorithms properly converge to the desired result with limited distortions. Additionally, we focus our attention on enhancing performance through data parallelism, achieved by increasing parallel memories.This adaptation enables tuning the architecture for the desired computation performance. Data parallelism is a technique that involves processing large datasets simultaneously using multiple processing units.Parallel memories, on the other hand, refer to a memory architecture that enables multiple memory modules to work in parallel.To achieve efficient data processing, parallel memories play a vital role as they enable concurrent data access.In this way, the memory system can handle multiple data access requests simultaneously, reducing potential bottlenecks and significantly improving the overall computation performance. Spectrum inversion using FFT The FFT reconstruction addresses the issue of recovering an estimation of the input x from regularly sampled interferograms as obtained in eq.(6b).Despite only working under restricting conditions, it is vastly more efficient to use a FFT than a generalized discrete Fourier transform (DFT) in terms of hardware implementation. The FFT requires to respect a certain order of calculations, since the data at a certain stage depends on previous computations.The Cooley-Tukey radix-2 [4] has proven to be a very popular choice to implement the FFT algorithm, as its memory-based architecture allows to freely choose the order of execution.In this work, we use the BFP as data format for improved precision in data representation. We propose to enhance the existing Cooley-Tukey implementation by introducing a pre-and postprocessing normalization step for each computational block.The main idea of this proposal is to share the exponent of the BFP precision between FFT stages, eliminating the drawbacks of the normalization block.This modification, implemented within the constraints of HLS which impose to follow specific design templates, results in accelerated calculations about 3 times faster than the classic implementation. We firstly introduce the BFP precision in Section 3.1.1and we showcase the proposed modifications to the Cooley-Tukey implementation in Section 3.1.2,detailing its pre-and post-processing normalization steps. Block floating point (BFP) In this paper we employ BFP as a data format.This choice allows for improved precision and efficiency in representing the data, as it combines the benefits of fixed point (FxP) and floating point (FP) data formats.The BFP is a variant of FxP format that includes an exponent component within a block of N values (x 0 , x 1 , ...x N −1 ).The shared exponent determines a scale for the normalized block (x ′ 0 , x ′ 1 , ... x ′ N −1 ), similar to how FP handles each individual variable [19]: In the case of FFT, a block consists of all the values calculated in a single stage.Therefore, the inputs of an FFT stage are scaled based on the maximum value from the previous stage.The position of the leading bit in the maximum value determines the magnitude of the scale accordingly. Subsequently, the next input block is shifted according to the leading bit position helping to prevent overflow while the overall shift value is recorded in the exponent component γ.Ultimately, the correct scale is derived from the shared exponent in eq. ( 15).Consequently, BFP avoids unnecessary truncation, resulting in an improved dynamic range, while maintaining lower computational overhead [20,26]. Proposed inversion architecture The implemented memory-based architecture is illustrated in Figure 2. It features a single butterfly and sequential operations, with required data loaded from memory as the computation progresses.Multi-bank memory is used to enable the calculation of all branches of the butterfly in parallel and the data are initialized in a way that avoids conflicts and then enhance parallelism.In accordance with the FFT graph [17], preand post-butterflies swapping schemes are necessary to maintain data separation.Failure to do so after the first FFT stage can result in required operands being placed in the same memory bank, leading to improper operation.The proposed design incorporates a single butterfly and memory structure with multiple banks, enabling the complex data operands to be loaded in parallel as computation is progressing.Fig. 2: FFT computation structure using parallel memories.The presence of R memory banks enables the calof a radix-R FFT complex butterfly operation in a single clock cycle.For radix-2 scheme, these memory banks provide parallel access to the two complex operands required for each operation in the FFT stages. The consistency and regularity across all FFT stages facilitate to use optimization techniques such as pipelining and unrolling to achieve faster computation.Pipelining with instruction-level parallelism ensures continuous execution across blocks, starting new iterations before previous ones finish [29] while the initiation interval defines the timing interval between consecutive iterations.Both parameters control the scheduling during HLS generation.In the case of our design, the FFT loops are pipelined with an initiation interval of one clock cycle as there are enough ports in the memories to read the data in parallel.Furthermore, the remaining loops responsible for bit operations, including leading bit and address generator, are unrolled to maximize performance and enhance overall efficiency. Pre-butterfly normalization Pre-butterfly normalization shifts input data in stage T based on the leading bit computed at the end of stage T − 1.This dependency prevents a scheduling generation with a small initiation interval at high frequencies.With an initiation interval of 1, lower frequencies enable the generation of a scheduling that aligns the leading bit and normalization block within a single cycle, ensuring correct operation.However, at higher operating frequencies, delays in butterfly arithmetic operations exceed a single clock cycle.This delay results in having the leading bit at least one clock cycle ahead of the normalization block in stage T which creates a data feedback loop, hindering the correct schedule generation by HLS tool.One solution is to increase the initiation interval beyond the feedback delay, which can resolve scheduling problems.However, this extends the computation cycle count. Post-butterfly normalization The proposed postbutterfly normalization directly shifts the outputs of the butterfly to the next This modification has an impact on accuracy, as the effect of bit growth in a given stage is corrected at the end of the following stage.To ensure proper operation, it is necessary to select the number of bits for the integer part that can accommodate the worst-case bit growth between two successive stages.Theoretically, for radix-2, the worst-case bit growth is 4 bits.However, in practice, considering the arithmetic growth factor of only 2.41 for one stage as detailed in [6], the expected worst-case bit growth between two successive stages is reduced to only 3 bits. Post-butterfly normalization allows for maximum performance and is hence we propose it as the preferred choice for implementation in this paper. Spectrum inversion using matrix inversion In this section, we present hardware strategies for efficient matrix multiplication-based reconstruction.We also explore parallelism enhancements in hardware, focusing on their potential to significantly accelerate computations. Pseudo-inversion (PINV) approach The PINV strategy involves an inversion without regularization and whose closed form expression is given in eq.(10). The implementation involves matrix multiplication, with the pseudo-inverse matrix A † ∈ R N ×M and the interferogram vector serving as inputs requiring to conduct N × M multiplications.To illustrate this process, a simplified structure of a single multiplier is depicted in Figure 3.In hardware implementation, additional operations such as memory read and memory write are involved and must be considered in the scheduling process, resulting in a latency for the generated architecture that exceeds the N × M theoretical latency.Pipelining, unrolling and data parallelism can help to significantly alleviate the latency.Multiple parallel line-column multiplications can drastically speed up computation since each output element's multiplication is independent.Yet, using just one memory with the inversion matrix (Figure 3) is not feasible due to the two memory port limitation.To address this, we can split the rows of the matrix A † into K subsets and store them in separate memories (Figure 4).This allows to accelerate the computation by a factor of K compared to the single-memory setup.This new structure presents K inversions, each with a transfer matrix having (N/K) rows and M columns.Hardware optimizations from the single memory version apply to theses new inversions as well.The parameter K controls the computational efficiency of the produced implementations and enables assessment of both parallelism's advantages and its effect on hardware resources. Penalized SVD approach As discussed in Section 2.3, both the TSVD and TIK methods are inversion techniques consisting in penalizing the singular values contained in the diagonal matrix Ξ ′ obtained by SVD of the pseudo-inverse A † , as shown in eq. ( 13).We discuss in this section their hardware implementation. In practical terms, eq. ( 13) is composed by three matrix-vector multiplications that share the same computation scheme: The term O 1 is independent on y, so it is not required to be computed every time.However, the regularization parameter (i.e., R ′ for the TSVD and λ for the TIK) may change according to the specific conditions of the acquisitions and across different pixels.We hence implement the worst case scenario in order to be able to keep the adaptability of the parameters for every acquisition.Therefore, the implemented calculation structure is using multiple, physically distinct memories so that the parallelism can be applied through multiple memories for each matrix multiplication, as illustrated in Figure 5. For a matrix of size N × M , the total number of multiplication-addition operations in SVD-based inversions is given by R ′ (2N + M ), where R ′ denotes the number of singular values considered in truncation.In case of TIK, R ′ is equal the total number of singular values.Further improvements in latency can be achieved by introducing more parallelism, similar to the PINV method.The parallelism parameter K remains the same for all the multiplication blocks, and the structure is similar for both the TSVD and TIK inversion methods. Results and discussion The purpose of this section is to conduct a comprehensive evaluation of hardware implementation strategies for reconstruction.The focus will be on assessing inversion speed, parallelism, and reconstruction quality, in order to gain valuable insights into the performance and effectiveness of these strategies.We provide a brief summary of the experimental setup, highlighting the key components and configurations used.Subsequently, the discussion focuses on the resource overhead associated with the implementation and the inversion speed of the considered inversion architectures.After that, the role of parallelism is investigated to assess its impact on computation speed and resource utilization for each of the considered methods.Finally, the achieved reconstruction quality is evaluated, providing insights into the accuracy of reconstruction based on the method and architecture parameters.These analyses aim at providing a comprehensive understanding of hardware implementation strategies for reconstruction and present informed observations and conclusions regarding their performance. Experimental setup All the considered inversion algorithms have been implemented for Xilinx Zynq SoC FPGA.The experiments have been carried out on a Zybo Z7 dev board, which is equipped with a Xilinx Zynq-7020 SoC (XC7Z020-1CLG400C).The board includes an FPGA Fabric with 85,000 logic elements, 220 DSP slices and 4.9 Mb of memory blocks.Furthermore the chip includes some software/Hardware co-design elements such as a 667 MHz ARM Cortex-A9 processor and 1GB DDR3 RAM.The considered implementations lied purely on the FPGA fabric.The RTL has been generated with Catapult HLS IDE while the synthesis have been performed on Vivado for the selected part. Discussion on resource overheads The comparison of hardware resources is presented in Figure 6a.Considering DSP blocks without parallelism, the matrix multiplication-based methods (PINV, TSVD, and TIK) exhibit lower overhead.The PINV method requires a single multiplier for the matrix-vector multiplication, while the SVD-based methods (TSVD and TIK) utilize two multipliers for the VΞ ′ and U T y multiplications.These multiplications are computed sequentially, using one multiplier for each operation.On the other hand, the FFT method requires 5 multipliers since all the operations are performed in a complex form. In terms of memory usage, the FFT method proves to be more advantageous with the lowest overhead.Only the twiddle factors and temporary data need to be stored in memory for this method.In contrast, the matrix multiplication-based methods require a larger number of memory blocks.Specifically, the PINV method requires memory to store the inversion matrix, while the TSVD/TIK methods require memory for storing the elements of the SVD factorization. Discussion on inversion speed Computation speed is achieved by dividing the operation count by the maximum operating frequency.Parallelism is dictated by available memory ports, and it affects the speed through the parallelism factor K. Figure 6b illustrates operating performances.Among methods, the FFT exhibits the swiftest performance, measured as clock cycles divided by max frequency.This surpasses non-parallel matrix multiplication approaches. It is approximately 8 times faster than PINV and 24 times faster than TIK with a single memory (Figure 6b).These improved performances can be attributed to the reduced redundancy in the FFT algorithm.However, the speed difference can be mitigated by utilizing data parallelism.For example, by employing 6 memories in parallel for PINV and TIK, the processing speed is only 1.5 and 3.2 times slower, respectively, compared to FFT.Further reduction in the speed difference can be achieved by increasing the number of memories, albeit at the expense of additional DSP blocks.In the case of PINV, the number of DSP blocks increases proportionally to the number of memories, while this rate doubles for TIK. The reported speed for TSVD corresponds to the worst case scenario where all the singular values are considered.However, when a lower number of singular values is used, TSVD can achieve improved inversion speed compared to PINV.On the other hand, TIK consistently performs 4 times slower than PINV since all the singular values are utilized in the calculation. Discussion on parallelism in TSVD/TIK As discussed earlier, both TSVD and TIK methods share the same computing architecture.Unlike the PINV method, this architecture requires additional storage capacity for the three matrices.The overhead resulting from this storage requirement is evident in Figure 6a, where the number of memory blocks used is nearly three times higher compared to PINV.This overhead remains relatively constant regardless of the parallelism factor.It is worth noting that the relationship between computational latency and the parallelism factor follows a similar trend as in the case of PINV, with a latency factor of 5.5 when the parallelism factor ranges from 1 to 6. From a hardware overhead perspective, PINV clearly offers advantages.However, the subsequent discussion will demonstrate that this may not be the case when considering the reconstruction quality. Discussion on the reconstruction quality This section focuses on reconstruction quality from different inversion methods, shown in Figure 8a.FFT does not guarantee superior quality even with precise interferograms.On the other hand, both PINV and TSVD methods demonstrate better overall quality, with comparable results regarding the required precision of interferograms to achieve maximum quality.This can be attributed to the fact that PINV is a special case of TSVD, where all singular values are considered. To achieve higher quality reconstructions, the TIK method proves advantageous as it offers significant improvements over other matrix multiplication-based methods.However, to achieve the same maximum quality as PINV/TSVD, TIK requires additional bits (8 bits for PINV/TSVD vs 12 bits for TIK).Nevertheless, with 4 bits precision increase, TIK enables a significant improvement in quality, surpassing the other methods based on matrix multiplication by approximately 30%.Overall, these findings emphasize the importance of considering reconstruction quality alongside speed and hardware resources when selecting an inversion method. Conclusion In this paper, we conducted a comprehensive evaluation of hardware implementation strategies for spectrum reconstruction from interferograms acquired by FTS.Specifically, this study focused on inversion speed, parallelism, and reconstruction quality.The resource overhead analysis highlighted the advantages of matrix multiplication-based approaches in terms of DSP block utilization and memory requirements.The investigation into inversion speed demonstrated improved efficiency for the FFT method while parallel matrix multiplication methods exhibited competitive performance at the cost of higher resource utilization. Considering the reconstruction quality, the FFT exhibits the lowest reconstruction quality, even with precise interferograms, compared to PINV and TSVD.Notably, the TIK method showed promising potential for achieving higher quality reconstructions when considering a slight increase in bit precision.Although TSVD/TIK have architectural similarities, the reconstruction quality assessment revealed nuanced differences between the methods, emphasizing the significance of considering precision requirements and bit-depth when targeting higher-quality results. By providing a comprehensive understanding of these implementation strategies, this paper intends to facilitate informed decision-making in selecting the most suitable approach for specific applications and target hardware. Fig. 7 : Fig. 7: Hardware overhead comparison in terms of parallel memories. Fixed point inversion comparison.	2023-10-04T06:42:14.235Z	2023-10-03T00:00:00.000	{ "year": 2023, "sha1": "da6df1564a6706ad2ef96138b481b44dc93b8d2f", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "da6df1564a6706ad2ef96138b481b44dc93b8d2f", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Engineering" ] }
1919087	pes2o/s2orc	v3-fos-license	VisHiC—hierarchical functional enrichment analysis of microarray data Measuring gene expression levels with microarrays is one of the key technologies of modern genomics. Clustering of microarray data is an important application, as genes with similar expression profiles may be regulated by common pathways and involved in related functions. Gene Ontology (GO) analysis and visualization allows researchers to study the biological context of discovered clusters and characterize genes with previously unknown functions. We present VisHiC (Visualization of Hierarchical Clustering), a web server for clustering and compact visualization of gene expression data combined with automated function enrichment analysis. The main output of the analysis is a dendrogram and visual heatmap of the expression matrix that highlights biologically relevant clusters based on enriched GO terms, pathways and regulatory motifs. Clusters with most significant enrichments are contracted in the final visualization, while less relevant parts are hidden altogether. Such a dense representation of microarray data gives a quick global overview of thousands of transcripts in many conditions and provides a good starting point for further analysis. VisHiC is freely available at http://biit.cs.ut.ee/vishic. INTRODUCTION Microarrays have become the standard way of producing genome-scale measurements of gene expression levels (1). Since the first experimental studies (2), microarrays have been used for answering a large variety of questions, such as characterizing gene expression patterns in tumour cell lines and healthy tissues (3,4), identifying key mechanisms of stem cell differentiation (5), and reconstructing global transcriptional networks in model organisms (6). Databases like ArrayExpress and GEO (7,8) have become goldmines of transcriptomic information with thousands of publicly available microarray datasets. Interpretation and visualization is a crucial step of microarray analysis, as measurements are abundant and the level of experimental noise is high (9). A common reasoning behind microarray analysis is 'guilt by association', as genes with similar expression profiles may have common regulatory circuits and functions (10). Unsupervised clustering presented as a heatmap and dendrogram is a common approach for detecting coexpressed groups of genes (11,12). Gene Ontology (GO) annotations are often used for the biological interpretation of detected clusters (13). Clustering has several well-identified drawbacks that affect interpretation and reproducibility (14). Popular clustering methods rely on input parameters, for example, hierarchical clustering (11) applies a fixed dendrogram cut-off value, and K-means (12) require predefining the number (and hence, the structure) of expected groups. Enrichment tools that relate gene groups to GO categories need to be accessed separately, which complicates the analysis of hundreds of clusters. Analysing results of hierarchical clustering is complicated, since each node of the dendrogram represents a potential cluster. Moreover, given the hundreds of potentially relevant datasets in public databases, the manual work would be unreasonable. Data visualization is also technically challenging, since heatmaps with thousands of transcripts hardly fit on computer screens. These problems are still commonly tackled with ad hoc means, e.g. removing genes that are 'not interesting' due to constant expression levels. VisHiC (Visualization of Hierarchical Clustering) is a web server for analysis of gene expression data, that provides agile all-in-one service for hierarchical clustering, functional enrichment analysis and visualization. The tool provides a global overview of a given expression matrix and highlights its most significant functional aspects using GO analysis. VisHiC builds a compact clustering using functional enrichments rather than fixed user-defined thresholds, by pruning clusters where no enrichments are found. GO enrichment analysis is a common measure of gene cluster interpretation and a wide range of related tools has been created in recent years (15)(16)(17). Several microarray analysis pipelines are available, notably Expression Profiler (18), GeneXPress (19) and AMEN (20) incorporate clustering methods with downstream analysis of annotations, sequence information and protein-protein interactions. The ambiguity of clustering methods has created a need for algorithms that assess multiple clusters (21). Some previously published tools also use functional information for clustering (22)(23)(24)(25). More recently, Ovaska et al. (26) combine clustering of genes based on semantic similarity of GO with heatmap visualization. However, the above comprise downloadable software that require additional data and expensive local computations. Our web server, on the other hand, provides the latest information from public databases and uses speedoptimized algorithms of HappieClust (27) and g:Profiler (16) to provide fast clustering and functional profiling even for larger datasets. In conclusion, we believe that our server provides an enhanced and useful service to the community. THE VisHiC SERVER VisHiC (http://biit.cs.ut.ee/vishic, Figure 1) is a web server for integrated cluster analysis, interpretation and visualization of microarray data that: (1) performs a fast approximate hierarchical clustering of a user-provided gene expression dataset; (2) computes functional enrichments of all discovered clusters using GO, pathways and regulatory motifs; (3) creates a compact heatmap dendrogram of the expression dataset, revealing most important functional enrichments and hiding poorly annotated expression profiles. The input of VisHiC is a gene expression matrix in plain tab-delimited or Gene Expression Omnibus SOFT format. Alternatively, one may use an expression matrix from our selection of example datasets. VisHiC supports a wide variety of gene, protein and probeset identifiers for human as well as most eukaryotic model organisms. The output of VisHiC is a compact gene expression matrix represented as a heatmap dendrogram, similar to the format used in many gene expression analysis applications. The analysis consists of three consecutive steps as described below. Novel approximate algorithm allows rapid hierarchical clustering of gene expression data The first stage of VisHiC analysis involves clustering of the input gene expression matrix. Agglomerative hierarchical clustering (AHC) organizes the data into a dendrogram, i.e. a tree where every node represents a gene cluster (28). Nodes in the bottom of the hierarchy (i.e. leaf nodes) represent single-gene clusters, all nodes except leaves are made up of two smaller clusters, and the root node contains all genes in the dataset. The AHC algorithm starts from single-gene clusters, iteratively merges most similar neighbours and results in a hierarchical structure of N À 1 non-trivial clusters given a dataset of N genes. Computational speed is an important consideration of AHC, as the standard algorithm requires all pairwise distances between expression profiles. This renders to around 200 million distances in case of an average mammalian genome. The VisHiC server incorporates HappieClust, our novel approximate version of the AHC algorithm (27). Instead of computing all pairwise distances, HappieClust takes advantage of pivot-based similarity heuristics to calculate all distances between similarly expressed genes as well as a random subset of more distant pairs. Since only a subset of all pairwise distances is calculated, HappieClust approximates the full AHC based on the pairwise distances that have been calculated during the process. Computational experiments with public microarray data show that HappieClust produces a biologically comparable analysis an order of magnitude faster than standard AHC. Pearson correlation is the default measure in VisHiC for determining similarity between expression profiles. Alternatively, one may apply the negative correlation measure that detects inverse correlation patterns such as those shared by a repressor and its targets. Absolute correlation is a combination of the two, as it detects both direct and inverse similarity. Functional enrichment analysis reveals optimal gene clusters of biological relevance The second stage of VisHiC analysis involves functional enrichment analysis of all detected clusters to infer the optimal clustering. A common strategy for partitioning a hierarchical clustering involves a dendrogram cut-off. However, it is difficult to provide a biologically plausible cut-off value, as gene expression profiles are not uniformly distributed and a fixed cut-off for different datasets does not guarantee stability. In this work, we take a different approach and infer clusters using statistical analysis of functional annotations [refer to (29) for a relevant review]. We use our g:Profiler software (16) to profile all discovered clusters for GO terms (13), pathways of Reactome and KEGG (30,31), regulatory motifs of Transfac (32) and microRNA target sites of miRBase (33). VisHiC applies the cumulative hypergeometric test to detect the significance of a functional annotation a, given that there are k genes in a cluster of n genes with an annotation a, and there are K annotated genes among the total of N genes in the genome: To evaluate the total enrichment in a given cluster, VisHiC computes a size-weighted annotation score q that summarizes enrichments of GO as well as pathways and regulatory motifs: Alternatively, one may opt for a strategy that assigns the best log P-value to each cluster, giving more preference to clusters with specific annotations: In order to reduce the amount of false positives resulting from numerous enrichment tests, VisHiC computes a special multiple testing correction that accounts for the hierarchical structure of GO (34). Standard corrections such as Bonferroni and Benjamini-Hochberg False Discovery Rate are also applicable. Enrichment-driven pruning of clustering dendrogram creates a compact view of expression data The final stage of VisHiC analysis creates a compact and biologically motivated clustering of the expression dataset to reveal its functional essence. Hierarchical clustering places gene groups in a parentchild structure, where clusters up in the hierarchy naturally contain smaller clusters as subsets. Similarly, the GO comprises a structured vocabulary where smaller groups of specific annotations are contained in large general groups. Hence, one expects to see specific enrichments in child clusters and corresponding general annotations in parent clusters. As the clustering dendrogram contains a spectrum of hierarchically contained clusters from single genes to the whole genome, choosing an optimal cluster involves maximizing certain criteria within a branch. We have devised the following two-stage greedy algorithm that determines the cluster structure based on functional annotations. First, we look for dense clusters, i.e. clusters with a high annotation score q, or alternatively, the term with the strongest P-value m. We scan all groups of genes that have functional enrichments, greedily starting from the one that provides the strongest annotation score. A cluster is not considered if any of its child or parent clusters is already a dense cluster. Dense clusters are shown in the final output. Second, we detect sparse clusters, i.e. groups of genes that have poor or no functional enrichments. We start the analysis from the root of the dendrogram and pass it recursively, compressing all clusters except the ones that contain dense clusters as child nodes. Sparse clusters are cut-off from the dendrogram and corresponding expression profiles are hidden in the heatmap. Our annotation-driven clustering algorithm is fully automated and does not depend on user-defined cut-offs. Cluster boundaries are determined only from significant enrichments of functional terms. VisHiC excludes small (<5 genes) and large (>1000 genes) clusters from enrichment analysis for optimal running time. The user may choose a different range of cluster sizes, or disable all compression to view the full expression matrix with all related enrichments. All functional terms that remain significant after multiple testing correction are used for computing the optimal clustering. However, one may apply a more stringent P-value threshold to reduce the number of contributing enrichments and compress the matrix to a greater extent. The resulting expression matrix is presented as a heatmap of gene activation and repression patterns, complete with a dendrogram that highlights functional groups of coexpressed genes. Colour-coded rectangles in the dendrogram denote dense clusters and related functional categories (GO, KEGG, Reactome, regulatory motifs, microRNA target sites). Cluster-specific functional annotations are additionally presented in a table and also appear when hovering over the dendrogram. The main window displays the compact heatmap with all highlighted clusters, while one may also 'zoom in' to view any cluster separately. In compact view, vertical branch stumps of the dendrogram mark places where sparse clusters are compressed. The user may search for genes of interest, or conduct further analysis via hyperlinks to external resources, e.g. browse-related functional categories via the GO web site or g:Profiler. Results: expression profiles of heart tissue of cardiovascular patients contain clusters related to muscle, mitochondria and extracellular matrix We present a case study to demonstrate the use of VisHiC in biological analyses (Figure 1). The example comprises a microarray dataset of myocardial remodelling, including 38 samples from 3 clinical groups of patients with ischemic, non-ischemic and myocardial infarction, taken before and after left ventricular assist device implantation [available in GEO as part of the series GSE974 (35)]. We clustered the dataset, detected optimal clusters with best enrichments and visualized the resulting expression matrix (Figure 1a). We used a custom stringent P-value threshold (P < 10 À7 ) and 'best annotation' cluster selection strategy with Pearson correlation measure to compress the matrix into a reasonable publication-sized format. The best scoring clusters are related to mitochondrion (Figure 1b), muscle tissue ( Figure 1c) and extracellular matrix, all of which are expected to be present in heart tissue expression profiles. Mitochondria produce adenosine triphosphate (ATP) and are the primary cellular energy generators. A recent publication underlines the importance of mitochondria in the heart and relates its mutations to heart disorders (36). The cluster with muscle tissue enrichments (ID:36899, see Figure 1e for expression profiles and Figure 1f for functional annotations) contains 420 probesets for 251 genes and has several strong enrichments (contractile fibre: P < 10 À28 , muscle system process: P < 10 À22 , cytoskeletal protein binding: P < 10 À19 ). In addition, our analysis reveals an enrichment for the binding site of serum response factor (SRF) (Transfac M01007, P < 10 À9 ). SRF is a known heart transcription factor which increased expression in congestive heart failure (37). The case study shows that VisHiC successfully extracts relevant functional aspects of a dataset, and compresses it into an easily perceivable compact format that fits well on screen and paper. DISCUSSION AND CONCLUSION VisHiC (http://biit.cs.ut.ee/vishic/) is a public web server for clustering and interpreting gene expression data. The tool is designed to extract the most significant biological features of a microarray dataset in a single run. The main output is a compact global view of the expression matrix with only the most significant clusters shown and less pronounced patterns hidden away, as its interactive format leaves open ends for more detailed analyses. VisHiC provides stability to otherwise ambiguous clustering and performs the labour-intensive task of evaluating hundreds of redundant clusters in a rapid automated manner. The approximate hierarchical clustering and rapid functional analysis guarantee meaningful results even if the datasets are large. Functional assessment of microarray datasets is an immediate application of VisHiC analysis, as annotations of highlighted clusters should relate to proposed hypotheses. Our approach is likely to be useful for large expression data warehouses, so that first broad overviews could be offered to users who are routinely browsing hundreds of datasets. One may use VisHiC to compare different datasets in the context of experimental conditions, global expression patterns and functional aspects. Integrating expression clusters with other types of experimental data like protein-DNA and protein-protein interactions may provide researchers with additional clues about gene regulation.	2014-10-01T00:00:00.000Z	2009-05-29T00:00:00.000	{ "year": 2009, "sha1": "c322f431638b2ef62962ec6e64b26d17c3f9766e", "oa_license": "CCBYNC", "oa_url": "https://academic.oup.com/nar/article-pdf/37/suppl_2/W587/3962499/gkp435.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "c322f431638b2ef62962ec6e64b26d17c3f9766e", "s2fieldsofstudy": [ "Biology", "Computer Science" ], "extfieldsofstudy": [ "Medicine", "Biology", "Computer Science" ] }
141166974	pes2o/s2orc	v3-fos-license	tilburg Using Oral History Methods to Document the Subjective Experiences of Statelessness The Case of Stateless Chinese-Bruneian Immigrants in Vancouver This article reflects upon the use of oral histories in uncovering people’s subjective experiences of statelessness — an area that has received relatively little attention in the growing body of literature on statelessness. Through an analysis of 13 oral history interviews with formerly stateless Chinese-Bruneian immigrants living in Vancouver, this study sought to understand the emotional and material repercussions of being denied a nationality, as well as respondents’ conceptions of citizenship and civic behaviour. By privileging the voices of formerly stateless people and giving them the opportunity to tell their life stories using their own words, this study advocates for the need to pay greater attention to the subjective, quotidian dimensions of this global human rights crisis. Introduction What is it like to be stateless? How does being stateless affect people's everyday lives? How do stateless people make sense of concepts like national identity and belonging? To probe into these questions, I decided to delve into the lives of a population that has hitherto received no academic attention: stateless Chinese-Bruneians, many of whom have left their homeland and have come to settle in Vancouver, Canada. Soon after I began this inquiry, however, I came to the realization that the difficulty did not just lie in demonstrating the academic value of understanding people's subjective experiences of being stateless. Nor was I just confronted with the challenge of building upon the very little information that exists about the ethno-political history of Brunei as a whole, much less the conditions of a minority group whose very legal existence has largely gone unrecognized. In addition to these predicted difficulties, I discovered that many of the respondents themselves simply did not believe that their own stories were worthy of being heard. My requests to speak with them about their pasts were often met with incredulity or skepticism: 'Why do you want to study such a small country?'; 'I left [Brunei] such a long time ago! What does it matter anymore?'; 'Who would care about Chinese people there?' While these reactions may partly be attributable to shyness or modesty about divulging personal details in an interview, more profound factors may have played a part in producing the self-effacing and dismissive attitudes of many of these respondents. Thus, the outcomes of this research grew to being twofold in nature. Firstly, I sought to humanize the issue of statelessness by describing its emotional and material impacts on Chinese-Bruneian immigrants in Vancouver, and by showing how Brunei's and Canada's citizenship policies differentially mediated their civic attitudes. Secondly, by providing an opportunity for people to tell their life stories using their own words, this project served as a way for stateless Chinese-Bruneians to revalue their own histories, and to raise greater awareness about a crisis that has largely remained invisible to the public eye. A Brief Background on Stateless Chinese-Bruneian Immigrants in Vancouver The people upon whom this study focused originated from Brunei Darussalam, a small oil-rich hereditary monarchy that gained independence from Britain in 1984. Covering a relatively tiny area of 5765 square kilometers on the northern end of Borneo island in Southeast Asia, it currently has a population of 402,000 comprising several ethnicities. This includes the dominant Malay group, which forms just over two-thirds of the country's inhabitants, with the Chinese constituting the largest minority group at around 11% of the population, according to 2011 estimates.1 Despite its multiethnic makeup, citizenship in Brunei Cheong tilburg law review 19 (2014) 74-80 functions as a mechanism by which the monarchy protects the cultural, economic, and religious interests of the dominant Malay majority, often to the detriment of other groups. Those who fall outside of this category are not automatically entitled to citizenship, regardless of whether they were born on Bruneian soil. Instead, they are given stateless permanent residence status, and are issued International Certificates of Identity in lieu of Bruneian passports.2 Like many other stateless peoples around the world, they are not only denied legal membership to the country in which they live, but they are also unable to make claims to fundamental civil, political, economic, or social rights. Confronted by these constraints on an everyday basis, stateless Chinese-Bruneians have increasingly turned to permanent out-migration as a strategy for responding to this institutionalized marginalization. Along with the United Kingdom and Australia, Canada has been a popular destination country. In the present day, there are 4825 Bruneian immigrants residing in Canada, most of whom are formerly stateless Chinese who have settled in either Vancouver or Edmonton.3 To echo one of the respondents: Why study a marginal minority of a faraway microstate? For one, systemic human rights violations should not be disregarded merely on the basis of the size of the populations affected. Secondly, despite being small in number, the case of Chinese-Bruneians who have moved to Canada is instructive for understanding the connections between citizenship and migration, precisely because citizenship (or more accurately, the lack thereof) figured so prominently in their everyday lives. By comparing the ways in which stateless Chinese-Bruneian immigrants talked about their experiences in Brunei and Canada, we can also make sense of how citizenship policies can affect people's interpretations of their relationships to the State, and how these interpretations can shift as they move from one national environment to another. In total, 13 semi-structured interviews were conducted with Chinese-Bruneians who had settled in Vancouver, Canada, from anywhere between 1 and 38 years. The majority had naturalized in 3 to 4 years of coming to Canada, with the exception of 2 people who took 6 years and 2 who had only recently arrived as permanent residents and were not yet eligible to apply for citizenship. There were 8 men and 5 women in total, ranging from 27 to 75 years of age. The oral history method was selected in order to provide a human face to a legal condition that has undergone little qualitative investigation. The conversations, which took place in participants' homes or in coffee shops, revolved around their memories of being raised in Brunei, the impacts statelessness had on their daily lives, their attitudes towards the Bruneian and Canadian governments and citizenship policies, their reasons for immigrating, and their experiences of integration into Canadian society. These dialogues were guided by the following questions: What are the material and subjective repercussions of being stateless in Brunei? How did statelessness factor in to respondents' strategic decisions to immigrate out? And lastly, how do the respective citizenship models of Brunei and Canada structure individuals' understandings of citizenship, civic identity, and national belonging?4 Through the privileging of stateless Chinese-Bruneian immigrants' voices, I argued that while Brunei's exclusive policies restricted people to an instrumental understanding of citizenship as restricting access to material entitlements, Canada's civic nationalist model guided respondents to espouse more symbolic ideals of freedom, democratic participation, and civic engagement. Bruneian Citizenship: Instrumental and Exclusive5 In recounting their earlier years in Brunei, respondents spoke of Bruneian nationality in primarily instrumental ways. The denial of legal status was perceived as a way for the government to restrict certain groups from enjoying an array of government privileges and benefits. As one respondent put it: Even though I was born in Brunei, I was never a 'Bruneian.' My status was a Brunei Permanent Resident, even though I was born there. It's a very weird thing because when I talk to people, they just don't understand. I identified 3 main factors contributing to respondents' instrumental attitude towards citizenship: the material constraints imposed on their life chances-particularly in the areas of education, employment, and property ownership; the lack of freedom of expression or governmental accountability; and the impossibility of fitting in to the state-defined idea of Bruneian nationality as tied to the Muslim Malay identity. Though the government-imposed restrictions which affected all aspects of people's lives, the state's emphasis on the selective withholding of material privileges allowed respondents to see themselves in a sense as free from obligation to the state. Instead of regarding being part of the Bruneian national community as performing one's civic duties in a mutually accountable individual-state relationship, respondents instead viewed citizenship as a mechanism by which the government restricted a unidirectional flow of material entitlements to the individual. Canadian Citizenship: Symbolic and Inclusive6 Confronted by these wide-ranging disadvantages stemming from Brunei's restrictive citizenship policies, many stateless Chinese turned to out-migration as a strategy for improving their life trajectories. When asked about their reasons for obtaining Canadian citizenship, however, respondents did not focus on the material benefits that were now open to them. Rather, I observed a marked shift in the ways in which respondents talked about Canadian citizenship and their roles as citizens in the Canadian context. As our conversations progressed to their experiences in Canada, respondents' associations with citizenship moved away from material entitlements and towards more symbolic ideas of democratic freedom, civic participation, and the chance to finally belong to a nation. For example, one respondent stated, '[s]o I think the biggest difference [between Brunei and Canada] is that here you can say you are a Canadian. And we are pretty proud once we have a country to call our own, because we were never wanted [in Brunei].' To account for this change, we must look to the broader institutional contexts in which these respondents' attitudes developed. Exposed to state-led discourses describing Canadian citizenship as entailing both rights and responsibilities, stateless Chinese-Bruneian immigrants reiterated ideals of civic participation and democratic freedom, and only considered matters of material benefits secondarily. In addition to celebrating their newfound sense of empowerment through the ability to vote and express their opinions on public issues affecting them, respondents also drew upon images of Canada as a proverbial land of liberty, egalitarianism, and multicultural rights. They saw citizenship not as a tool to exclude and stratify, but as a mutually obligatory relationship between individual and the state. This being said, more complicated dynamics underlay the apparent success stories of stateless Chinese-Bruneians finding new homes in Canada. I must point out the disjuncture that sometimes became apparent between these romanticized perceptions and the less than rosy realities of their integration experiences. This was most noticeable in respondents' accounts of trying to enter or move up in the Canadian labour market. While approximately one third of respondents managed to obtain white-collar careers and resided in middle-class neighbourhoods in Vancouver, others became stuck in lowpaid jobs traditionally held by immigrants and ethnic minorities, such as those in hospital cafeterias, hotels, and car-washes. Barriers such as the nonrecognition of foreign credentials, th e lack of financial resources to upgrade their skills, the privileging of 'Canadian work experience' , and possible stereotypical assumptions held by employers about hiring immigrants in blue-collar positions, posed significant challenges to respondents adjusting to their new lives. Despite these gaps between discourse and reality, respondents did not dwell upon these hardships. That they opted instead to highlight the more positive symbolic benefits of being part of the Canadian national community points to the power of institutionalized discourses over the way people make sense of their own circumstances. Conclusion This study marks the first attempt to critically analyze the subjective experiences of stateless Chinese-Bruneians and their attitudes towards citizenship. In addition to describing the oft-neglected human aspects of being denied the right to a nationality, findings from this project also contribute to our understanding of the interplay between citizenship policies and the ways in which individuals relate to the state.7 These oral histories of stateless Chinese-Bruneian immigrants in Vancouver shed light upon how formal laws and discourses surrounding what it means to be a citizen influence how individuals understand their own roles as civic beings, and how they fit themselves into their national communities. In contributing to the relatively nascent but growing body of literature on statelessness, I argue that more attention needs to be paid to human stories and the subjective, quotidian dimensions of this global human rights crisis.	2019-05-01T13:08:12.297Z	2014-01-01T00:00:00.000	{ "year": 2014, "sha1": "9871a59d514bce5318203b47d154d0ddb59e4511", "oa_license": null, "oa_url": "https://doi.org/10.1163/22112596-01902008", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "e73578b26590070da9c8be439fceef88f984d2e5", "s2fieldsofstudy": [ "Sociology" ], "extfieldsofstudy": [ "Sociology" ] }
133587007	pes2o/s2orc	v3-fos-license	Half-Life of Glyphosate on the Control of Water Hyacinths in Water Tanks The present study had the purpose to assess the behavior of the herbicide glyphosate in the control of water hyacinths, in a water environment through half-life, in a controlled and closed experimental field, in order to contribute to the analysis of the risk of the environmental impact of the use of this product. Eight fiber-cement tanks with the storage capacity of 250 liters were used under the conditions without water flow and without replacement of water. The experiment was designed in two treatments with four repetitions, the first being colonized by water hyacinths and the second without the presence of the macrophyte. The recommended dose, according to the manufacturer, was 7.0 L∙ha or 3402 g∙ha of equivalent acid. For the application in the tanks, we have used carbon dioxide (CO2) precision equipment, providing a flow rate of 200 L∙ha. The samples were collected on the day of the application (day zero) and 2, 4, 8, 16, 32 and 64 days thereafter (DAP). The method used for determination of residues was by high performance liquid chromatography (HPLC) and mass spectrometry with a mass selective detector. From the mathematical model of charge decay, the half-life of glyphosate in water was estimated to six days for the tanks without water hyacinths and sixteen days for the tanks colonized by macrophytes. The result obtained permits to conclude that the glyphosate in water is quickly degraded in closed water environments and does not leave residues that would prevent its use. Introduction The chemical method, which advocates the use of specific herbicides to control aquatic plants, has been the most widely used in different places in the world due to low cost and especially the speed, easy of application and control efficiency [1] [2] [3].The world's most used herbicides to control aquatic macrophytes are: 2,4-D; diquat; endothall, a copper-based compound; fluridone; imazapyr and glyphosate [4].The production of knowledge in this area in Brazil has been small, given the prohibitive legislation in loco studies [5].The studies developed in the country, reason of that prohibition, have been performed in controlled environments and in closed systems, with results showing herbicide use efficiency in controlling several species of aquatic plants [6] [7] [8] [9]. According to [5], the chemical control of aquatic plants is restricted to a few herbicides due to restrictions imposed by legislation, to the environmental impact, to the market size for the private sector and to the application of technology.The environmental restrictions are the most important factors because the waterways are used for many purposes, such as a source of water for human and animal consumption, crop irrigation, leisure activities, navigation and hydraulic energy generation. Although the different work done to prove efficiency of the use of various active ingredients in the control of aquatic macrophytes, few environmental impact studies have been conducted.Environmental impact studies have been little exploited to date due to the limitation of the use of chemicals in aquatic environments [10] for the protective intervention of IBAMA (Brazilian Institute of Environment and Renewable Natural Resources).However, much more important than the action of the product in the management of macrophytes is its direct action on integration with the environment in question.Among studies performed, the ones mentioned are those monitoring the effects of the herbicides on the quality of the water and the sediments [11] [12] [13] [14] and more rarely those that have investigated, besides the water quality parameters, the molecule residues in plants and in the water, in the reservoirs or mesocosms [13] [14] [15]. Thus, an accurate study of the use of herbicides to control the aquatic macrophytes in closed aquatic environment, from the point of view of environmental impact, may systematize data scientifically treated, and form knowledge to evaluate its behavior within the environment.Therefore, it will contribute in the definition of public policies by regulatory agencies in facing the serious problem represented by the uncontrolled proliferation of macrophytes in water bodies, particularly, by obstruction of the turbines in reservoirs for power generation.This fact has a direct effect on the production cost of electrical energy, which is essential for economic, technological and social development of the country. In the present experiment the herbicide glyphosate was chosen because it is registered in several countries for the control of aquatic macrophytes, being one of the main products studied, used in the world and with great potential in the management of aquatic weeds.Thus, it was chosen to study it in a controlled and closed experimental field, in the management of (Eichhorniacrassipes) in reservoirs.The behavior of the product's half-life was analyzed, aiming to contribute to the risk analysis formulation of environmental impact of the use of this product to control aquatic weeds. Materials and Methods The study was developed in the Experimental Area of the Núcleo de Pesquisas Avançadasem Matologia (NUPAM) belonging to the Faculdade de Ciências Agronômicas da UNESP (Department of Agronomic Sciences of the University of São Paulo), Botucatu (SP) campus, during the months from May to November 2011.The local climate, according to the Köppen method, is classified as humid mesotherm (Cfa), presents average temperature of the hottest month above 22˚C and small water deficiency between April and August.Eight (08) fiber-cement tanks were used, four for each treatment with a storage capacity of 250 liters, without water flow and without replacement of the evapotranspired water ("worst case").The water supplying the tanks came from the Companhia de Saneamento Básico do Estado de São Paulo (SABESP) (Sanitation Utility of the State of São Paulo).The samples were collected in the lower third of the tanks. Treatments The treatments used in the experiment were called: 1-tank without water hyacinths and with application of glyphosate; 2-tanks with water hyacinths and with application of glyphosate. For the set of conditions of the tanks containing macrophytes, the species analyzed was Eichhorniacrassipes (Mart.)Solms (water hyacinth).The plants were collected from the Tietê River, Botucatu region (SP) and placed in boxes to provide 90% occupancy of the tank surface. The herbicide control was carried out by application of glyphosate in the recommended maximum dose since the recommendations for the study of environmental impact are to look for the worst scenario "worst case".According to the former registration of the product, the Rodeo, in Brazil, the recommended dose was 7.0 L•ha −1 or 3402 g•ha −1 of acid equivalent.The surfactant Alterbane was added to the herbicide stock at a concentration of 0.5% (466.6 g•L −1 ) [16]. The tanks were sprayed, using the carbon dioxide precision equipment (backpack sprayer), provided with plywood bar containing two flat jet nozzles Jacto XR 110.02, operating at a pressure of 2 kgf•cm −2 (20 Kpa), providing an even distribution of the glyphosate in the tank and a solution consumption corresponding to 200 L•ha −1 .The calibration was performed on site based on the applicator speed in regards to the area worked. Evaluations The values obtained in the control efficiency of water hyacinth were submitted to variance analysis and to Tukey test, considering a 5% level of significance.The evaluations of effectiveness were carried out visually after 2, 4, 8, 16, 32 and 64 days after application (DAA), using percentage scale, where 0 (zero) represents no control and 100%, the total plant control [17]. The water samples to determine the glyphosate residues were collected on the day of the application (day zero) and on 2, 4, 8, 16, 32 and 64 DAA, in the morning, always at the same time, at 11 a.m.The depth of the collection was conducted between 15 -30 cm below the water level, and the samples were placed in plastic containers of 200 mL and stored in a freezer at −18˚C, in which the residue analyses were subsequently performed. The concentrations (mg•Kg −1 ) of glyphosate and AMPA (main metabolite of glyphosate) residues were determined.For determination of compounds in water, the samples were thawed, stirred, and filtered directly on Millex HV filter (Millipore) 0.45 μm, provided with a 13 mm durapore membrane, and placed in a 9 mm amber-colored vial (Flow Supply), with a capacity of 2 mL, for further quantification by LC-MS/MS. The molecular mass and fragments generated from each molecule are shown in Table 1.The fragment used for quantification of each of the compounds was always the first presented in Table 1, for each compound.The analytical curves for the compounds were constructed in the concentration ranges described in Table 2.The linear model used in the equations was quite adequate. The method used was by high performance liquid chromatography (HPLC) and mass spectrometer with a mass selective detector, presenting detection limit of compounds at 0.001 mg•kg −1 of the glyphosate. Before each collection, the volumes of the tanks were measured so that, later, the load (mass) of glyphosate could be calculated in time.The measurement was made with a ruler, considering the relationship between the height/volume of the initial mark of each tank and the height/volume of the mark corresponding to the level of the water at the time of the collection. The mathematical model of decay was applied to describe the decrease in the concentration of glyphosate and loads in water and to determine the half-life of the compounds.In this experiment two tanks, repetition 1 (No 1) of the treatment plant with glyphosate application and repetition 1 (No 5) without treatment plant application of glyphosate, had to be discarded.The reason was the failure to identify residues in the 1st analysis, immediately after the application, to the tank N˚ 1 and small concentration to the tank N˚ 5; in the following analyses of this tank, no residues were detected.Since the calibration was carried out with water before application, a volume was certainly left in the system, between the bar and the spray nozzles.Moreover, as these tanks were the first to receive the herbicide application and since they were set in sequence in the experimental field, it is easy to understand why no residues were found. Results and Discussion The data obtained in the present experiment with the use of the herbicide glyphosate, in the evaluated dose, showed a percentage of average control of 91.3%, presenting excellent efficiency in the control of the water hyacinths, agreeing with previous works [6] [7] [18]. The values of the glyphosate residues (concentrations and mass) and tank volumes found from dates 0, 2, 4, 8, 16, 32 and 64 DAA, for both sets of tanks, are presented in Table 3.For those tanks which have not received plants, it can be verified that there was a reduction of loads in 99.5%, 90.9% and 99.3%, in boxes 2, 3 and 4, respectively.As for the tanks colonized by the water hyacinth, the decay in the boxes 2, 3 and 4 represented, respectively, 97.1%, 96.1% and 96.6%. Likewise the concentration, the average load of glyphosate found in the tanks was greater, sooner after the application, being four times greater where there was no presence of water hyacinth. However, over time, the mass found in the reservoirs without plants matched the one found in reservoirs containing plants.From 32 DAA onward, the loads found in the tanks without the hyacinth were lower compared to the tank containing plants.The results show that from the metabolism of glyphosate and subsequent control effectiveness or death of the plants onward, the process of decomposition of dead plants released the product into the water in the tanks keeping its content, even if low, but higher thanks that received direct application of the product. To calculate the AMPA mass in the water in the treatments with and without water hyacinths, the water volumes were used in the tanks in each moment of sample collections and the respective concentrations are presented in Table 4.In the analysis of AMPA parameter, it was expected to see the same behavior in the mass obtained for the glyphosate.However, as can be seen, the larger masses found in the treatment that received the direct application of the product without the presence of water hyacinth, lasted throughout the study period.This behavior suggests that the metabolism of AMPA through the bioaccumulation process is superior to glyphosate.It was not possible to compare results of glyphosate and AMPA loads with other studies, since these are lacking in the available literature. In order to adjust the data observed of glyphosate residues in water over time, it was used a mathematical model described by Equation (1): where: M t -glyphosate mass at moment t; M 0 -glyphosate mass at the initial time considered; k-decay coefficient and t-time. Applying the Naperian logarithm in the Equation (1), it is obtained: This equation represents a linear model where the coefficient (K) identifies the mass decay of glyphosate over time.Since three tanks were used in the experiment, the average load was calculated, at times considered of the samples, to determine the average decay value for glyphosate.In order to obtain this value, the logarithm (Ln) was applied to mass values observed (Figure 1) getting the decay value for the tanks without plants of 0.115 day −1 . From the value found in the coefficient of average load decay of glyphosate, average values of glyphosate were simulated using the mathematical model of Equation (1) (Figure 2).With this model, it was possible to estimate the half-life of glyphosate applied directly to the water, without the presence of water hyacinths.Considering the estimated average load, the half-life was six days.Likewise, the simulation was performed for the tanks receiving the colonization of water hyacinth.The value of the coefficient K and the observed and simulated values of the average load are shown in Figure 3 and Figure 4, respectively.For the conditions of the colonized tanks by water hyacinth, the decay value was calculated equal to 0.043 day −1 .From Equation (2), it was estimated the glyphosate half-life in water for the treatment with the macrophyte, resulting value of 16 days.The simulated values for the half-life of glyphosate in water, in both situations, are in accordance to [19] [20] [21] [22] which reports that the average life of the molecule in water can range from 1 to 51 days and to [23], which states that, depending on the conditions of the water body, particularly to those linked to the full microbe activity can vary between a few days to two weeks. Studies performed in a forest ecosystem [24] [25] have shown that glyphosate had quickly dissipated in the lake waters with many suspended sediments, with the half-life ranging from 1.5 and 11.2 days. Furthermore, in the study developed by [30], which was a river water body (open system), it was observed that the half-life of glyphosate in this environment had varied between 60 and 100 hours. [31] evaluated the environmental fate of glyphosate in the water-sediment system with focus on its microbial metabolization.The results found demonstrated the key role of sediments in its degradation.Also, Glyphosate was detected below detection limit in the water compartment at forty days. In compliance with [32], this is a low toxicity herbicide that quickly dissipates in the environment.[33] compliments reporting that low bioavailability of glyphosate in natural waters is due to its quick degradation and quick microbial decomposition. Conclusions The study has shown that glyphosate applied directly to the water surface or to controlling water hyacinth plants in closed aquatic environments is rapidly degraded and has no residues which would preclude its use.The average life of glyphosate in water for the control of macrophytes was estimated at 16 days and the decay calculated value was equal to 0.043 day −1 . Therefore, the use of glyphosate in the control of aquatic plants can be rec- Figure 1 . Figure 1.Logarithm of the observed values of average mass (mg) of glyphosate in the tanks with hyacinth and with application of glyphosate. Figure 2 . Figure 2. Observed and simulated values of average mass (mg) of glyphosate in the tanks with water hyacinth.Back symbols are values of average mass.Bars are values of average mass maximum and minimum.Dashed lines are simulated values of average mass. Figure 3 . Figure 3. Logarithm of the observed values of average mass (mg) of glyphosate in the tanks without water hyacinth and with application of glyphosate. Figure 4 . Figure 4. Observed and simulated values of average mass (mg) of glyphosate in the tanks without water hyacinth.Back symbols are values of average mass.Bars are values of average mass maximum and minimum.Dashed lines are simulated values of average mass. ommended for use in continuous flow aquatic environments, since it has more favorable characteristics to dissipate and degrade herbicides, due to being less drastic than the one evaluated in this work.the Weed Science Center (NUPAM) of the College of Agricultural Sciences São Paulo State University for generous support in this project. Table 1 . Molecular mass and secondary ions of the analyzed compounds. Table 2 . Analytical curves and concentration ranges for each compound analyzed. Table 4 . Values of AMPA mass (mg) in the water in the treatments, in each tank, in the treatments with and without water hyacinth from dates 0, 2, 4, 8, 16, 32 and 64 Days after Application (DAA).	2019-04-26T14:27:00.192Z	2017-04-17T00:00:00.000	{ "year": 2017, "sha1": "f56afec914ec49b842bcc97398d2b190c6c04096", "oa_license": "CCBY", "oa_url": "http://www.scirp.org/journal/PaperDownload.aspx?paperID=75536", "oa_status": "GOLD", "pdf_src": "ScienceParseMerged", "pdf_hash": "f56afec914ec49b842bcc97398d2b190c6c04096", "s2fieldsofstudy": [ "Environmental Science" ], "extfieldsofstudy": [ "Geology" ] }
258556603	pes2o/s2orc	v3-fos-license	Relation of digital arterial dysfunction to alternative frequency weightings of hand-transmitted vibration This study compared the relative performance of alternative frequency weightings of hand-transmitted vibration (HTV) to predict the extent of cold-induced vasoconstriction in the digital arteries of HTV workers. The cold response of digital arteries was related to measures of daily vibration exposure expressed in terms of r.m.s. acceleration magnitude normalised to an 8-h day, frequency weighted according to either the frequency weighting Wh defined in international standard ISO 5349-1:2001 (Ah(8) in ms−2 r.m.s.) or the hand-arm vascular frequency weighting Wp proposed in the ISO Technical Report 18570:2007 (Ap(8) in ms−2 r.m.s.). The measure of daily vibration exposure constructed with the frequency weighting Wp (Ap(8)) was a better predictor of the cold response of the digital arteries in the HTV workers than the metric derived from the conventional ISO frequency weighting Wh (Ah(8)). This finding suggests that a measure of daily vibration exposure constructed with the vascular weighting Wp, which gives more weight to intermediate- and high-frequency vibration (31.5−250 Hz), performed better for the prediction of cold induced digital arterial hyperresponsiveness than that obtained with the frequency weighting Wh recommended in ISO 5349-1 which gives more importance to lower frequency vibration (≤16 Hz). Introduction Experimental studies have shown that the response of finger circulation to hand-transmitted vibration (HTV) is frequency-dependent 1,2) .Vibration frequencies above 100 Hz can induce a stronger vasoconstriction than lower frequencies in either the human finger or animal models 3,4) .Several epidemiological studies have reported that occupational exposure to intermediate-and high-frequency vibration is associated with an increased risk of a second-ary form of Raynaud's phenomenon called vibration-induced white finger (VWF) 5) .These findings are in contrast with the frequency weighting W h recommended by the ISO 5349-1 standard (2001) 6) which gives more weight to lower frequency vibration (≤16 Hz) for the assessment of vibration-induced health disorders. Within the EU VIBRISKS project 7) , the Italian research team studied the application of a supplementary hand-arm vascular weighting (W p ) proposed in the ISO/TR 18570 (2017) 8) , which assigns more weight to intermediateand high-frequency vibration.Findings showed that W p performed better than the ISO W h for the prediction of the incidence of subjective symptoms of VWF in a cohort of HTV workers (Fig. 1). The aim of the present study was to compare the relative performance of the vibration metrics constructed with either the frequency weighting W h (ISO 5349-1:2001) 6) or the frequency weighting W p (ISO/TR 18570:2017) 8) to predict, in addition to VWF symptoms, the extent of cold-induced vasoconstriction in the digital arteries of the VIBRISKS workers. Subjects The Italian VIBRISKS cohort included 249 vibration exposed workers (215 forestry operators and 34 stone workers) and 138 control men employed at the same companies and unexposed to HTV.They were investigated at baseline and annually in the autumn-winter seasons over a three year follow-up period.Of the vibration exposed workers, 177 participated in three follow-ups, 36 in two follow-ups, and 36 in one follow-up survey.All vibration exposed workers continued to work with vibratory tools during the follow-up. The design of the VIBRISK prospective cohort study and the characteristics of the cohort workers have been described in details elsewhere 9) .Subjective symptoms of white finger were investigated according to the minimal requisites established at the Stockholm Workshop '94 10) ; in addition, colour photographs were used to compare the various degrees of whiteness, cyanosis or redness of the fingers and hands 11) .A clinical diagnosis of white finger was made when the subject, in addition to a reliable history of subjective symptoms, indicated the photographs displaying well-demarcated blanching of the fingers. The cold test The cold test was performed with the subject in a supine position after a rest period of 20-30 min in a laboratory room with an air temperature of 20-22°C.The cold test was carried out by means of a strain-gauge plethysmographic technique according to the method recommended in international standard ISO 14835-2:2005 12) .The percentage change in finger systolic blood pressure (FSBP) from 30 to 10°C (%FSBP 10°) in a test finger (FSBP t ), corrected for the change in systolic pressure in a reference finger (FSBP ref ) of the same hand, was calculated as: Arm systolic pressure (ASP) was measured by an auscultatory procedure with a mercury sphygmomanometer and a standard rubber cuff (12×23 cm).The digital pressure index (DPI) at 30 and 10°C (x°) was calculated as the ratio of FSBP in the test finger (FSBP tx°) to ASP x°: DPI x° = (FSBP tx° × 100)/ASP x° (%) To avoid nicotine-induced vasoconstrictive effects on the digital vessels, tobacco users refrained from smoking for at least 2 h before testing.The subjects were invited to abstain from alcoholic beverages since the evening preceding the day of the cold test.The cold test at the cross-sectional and follow-up surveys was performed by the same health personnel who used the same method and apparatus (Digitmatic 2000, Medimatic A/S, Copenhagen, Denmark).The subjects underwent the cold test approximately in the same calendar period ( ± 2 wk) and FSBPs were measured in the same test and reference fingers at the various investigations. Vibration exposure Vibration measurements were made on the tools used by the forestry workers (chain saws, brush saws) and the stone workers (grinders, polishers, inline hammers) according to the recommendations of international standard ISO 5349-1:2001 6) .Triaxial (x, y, z) vibration magnitudes were measured as r.m.s.accelerations over the frequency range 1-4,000 Hz using the frequency weightings W h and W p displayed in Fig. 1. The root-sum-of-squares (also called "vibration total value", a v ) of the r.m.s.accelerations of tool i frequency weighted according to W h or W p (W f ) for the x-, y-and zaxes was calculated as: The results of the measurements of the tool r.m.s.acceleration magnitudes weighted with either W h or W p are reported in detail in a previous paper 7) . Daily vibration exposure was expressed in terms of r.m.s.acceleration magnitude normalised to an 8-h day (A(8)), frequency weighted according to W h or W p (W f ): ¦ where a v is the vibration total value of the r.m.s.acceleration of tool i, T i is the duration of the i th operation with tool i in hours, and T 0 is the reference period of 8 h. Measurements of daily exposure time were carried out by supervisors who used a stopwatch method and recorded the contact time the hands of the operators were actually exposed to the vibration from each tool 9) . Data analysis The statistical analysis of data was performed with the (8) or A p (8) were estimated by maximum-likelihood random-effects regression models for repeated measures over the follow up period. The Bayesian information criterion (BIC) was used to compare the fit of the regression models including alternative measures of daily vibration exposure 13) .According to the strength of evidence rules for the difference (Δ) in BIC between models, ΔBIC 0−2 suggests no difference in the fit between models, ΔBIC 2−6 tends to give positive support for the model with the smaller BIC, ΔBIC 6−10 provides strong evidence for the model with the smaller BIC. Results The occurrence of symptoms of white finger over the study period was 7.2% (n=10) in the controls and 21.7% (n=54) in the HTV workers (17.7% (n=38) in the forestry workers; 47.1% (n=16) in the stone workers).At baseline, there were no significant differences in age and anthropometric characteristics between groups, while current smoking was more prevalent in the HTV workers affected with VWF (Table 1).Daily vibration exposure in terms of either A h (8) or A p (8) was significantly greater in the VWF workers than in the HTV workers with no vascular symptoms (p<0.001).The duration of exposure (yr) was comparable between the two HTV worker groups. Baseline FSBPs and ASPs at 30 °C were similar in the controls and the HTV workers with or without VWF symptoms (Table 2).The vasoconstrictor response of digital vessels to cold expressed as either %FSBP 10° or DPI 10° was stronger in the VWF workers than in the controls and the non-VWF workers (p<0.0001). The relation of cold test outcomes (%FSBP 10°, DPI 10°) to measures of daily vibration exposure was investigated by means of three models with different sets of explanatory variables: (i) Model 1: A h (8) 3 and 4). After excluding the controls from data analysis, in all models one unit of increase in A h (8) (1 ms −2 ) or A p (8) (10 ms −2 ) was significantly associated with an increase in the vasoconstrictor response of the digital vessels to cold, i.e. decrease in either %FSBP 10° (Table 3) or DPI 10° (Table 4).As expected, VWF symptoms were strongly related to the cold response of finger circulation.The BIC statistic suggests a better fit when A p (8) rather than A h (8) was included in the models as a predictor of coldinduced digital vasoconstriction (ΔBIC 6-7 for models with %FSBP 10° outcome, ΔBIC 8-9 for models with DPI 10° outcome).When data analysis was limited to VWF workers alone (n=54), after adjustment for the severity of VWF score 14) and other potential confounders, there was positive evidence that A p (8) performed better than A h (8) for the prediction of the cold test outcomes (ΔBIC 3-4 for models with either %FSBP 10° or DPI 10° outcomes, results not shown).Among the non-VWF workers (n=195), there was also positive evidence of a better fitting for the models including A p (8) compared to those with A h (8), (ΔBIC 3 for models with either %FSBP 10° or DPI 10° outcomes, results not shown). Discussion In this study, the peripheral vascular function in the fingers of HTV workers was assessed by measuring FSBP during cold provocation with a strain-gauge plethysmographic technique.The methods and procedures for the measurement of FSBP during finger cooling were in accordance with the recommendations of international standard ISO 14835-2:2005 12) .In our previous laboratory investigation of the cold response of digital arteries in a large sample of 874 male workers representative of nine groups occupationally exposed to HTV and in one control group of 455 healthy men, the sensitivity and specificity of the cold test for detecting digital arterial hyperresponsiveness to cold were 87% and 94%, respectively, for %FSBP 10°, and 80% and 94% for DPI 10°1 5) .The positive and negative predictive values of the cold test were 75% and 97%, respectively, for %FSBP 10°, and 73% and 96% for DPI 10°. The repeatability of the FSBP measurements during finger cooling in a sample of healthy men over a 5-d period was acceptable, the coefficient of variation averaging 6.2% for both %FSBP 10° (range 3.8%−9.2%)and DPI 10° (range 2.2%−11.6%).On a group basis, the cold test with FSBP measurements could differentiate between healthy controls and HTV workers, and within these latter FSBP indices (%FSBP 10° or DPI 10°) were significantly lower in the subjects affected with VWF than in those without vascular symptoms.These findings suggest that strain-gauge plethysmographic recording of FSBP during local cooling is a valid laboratory method for a quantitative evaluation of cold-induced arterial vasoconstriction in the fingers of HTV workers.In this study, the metric A p (8) performed better than A h (8) for the assessment of the vasoconstrictor effect of cold in the digital arteries of HTV workers.This is consistent with our previous epidemiological findings that A p (8) was a better predictor of the occurrence over time of VWF symptoms in the VIBRISKS cohort compared to the measure of daily vibration exposure A h (8) recommended by ISO 5349-1:2001 6,7) . Vascular investigations have revealed that acute exposures to vibration with equal frequency weighted acceleration magnitude can provoke stronger reduction in the blood flow of the human finger for frequencies between 31.5 and 250 Hz compared with vibration at 16 Hz 1) .The reduction of finger blood flow increased with increasing frequency both during and after the delivery of vibration stimuli.There is experimental and clinical evidence that exposure to vibration with a frequency of 125 Hz has a powerful constrictive effect on the digital vessels of the human fingers 2,16) .It has been suggested that 125 Hz vibration can elicit a sympathetic vasospastic reflex mediated by subcutaneous Pacinian receptors which are highly responsive to this frequency 16) .Biodynamic and mathematical models of the human finger have shown that the finger resonant frequencies vary from 100 to 250 Hz 4,17) .Exposure to this frequency range is associated with increased vibratory stress and enhanced local absorption of mechanical energy resulting in greater strain and risk of injury to the soft tissues of the fingers.In animal models, exposure to high frequency vibration (250 Hz) was found to induce both functional (increased oxidative stress) and structural (arterial remodelling and narrowing) changes in the ventral tail arteries of rats 4) .The findings of these biodynamic, pathophysiological and morphological investigations provide biological plausibility to the epidemiological evidence of an increased occurrence of vascular disorders in HTV workers operating power tools generating high frequency vibration. Overall, the present study and our previous epidemiological surveys suggest that the evaluation of vibration exposure by means of W p frequency weighting is more appropriate for the assessment and the prediction of subjective symptoms and objective signs of vibration related vascular disorders compared to the W h frequency weighting recommended by the current ISO 5349-1 standard 6) .W p assigns more weight to intermediate-and high-frequency vibration while W h does not reflect the frequency-dependence of the vascular response of the fingers to vibration and tends to overestimate the vascular effects of lower frequency vibration (≤16 Hz) 5,18) . It should be noted that the findings of this study apply only to vibration-induced vascular disorders and not to other adverse health effects (e.g.sensorineural and musculoskeletal) caused by occupational exposure to handtransmitted vibration.This notion is pointed out in the document ISO/TR 18570:2017 8) and confirmed in our previous epidemiological studies of either neurosensory symptoms or neck and upper limb musculoskeletal disorders in the Italian cohort of the VIBRISKS research 19,20) . The VIBRISKS prospective cohort study suffers from some limitations that deserve attention, such as, for instance, (i) vibration measurements limited to currently used tools, (ii) daily vibration exposure expressed according to an energy equivalent time dependency, (iii) possible recall or feedback bias for both outcomes and exposure variables, (iv) short duration of the follow-up time (three years), (v) pattern of missing data for dropping out of the study.These potential sources of uncertainty have been discussed in details in previous papers of the VIBRISKS study to which the reader may refer 7,9) . Conclusions In this study of the Italian VIBRISKS cohort, a measure of daily vibration exposure (A p (8)) constructed with a frequency weighting (A p ) that gives more weight to intermediate-and high-frequency vibration than the conventional ISO frequency weighting (A h ) performed better than the derived ISO metric A h (8) for the assessment of vibrationinduced peripheral circulatory dysfunction as revealed by an abnormal response of the digital arteries to finger cooling in the HTV workers.This is consistent with our previous epidemiological findings that A p (8) was a better predictor of the occurrence of VWF symptoms in the same cohort compared to the ISO A h (8) 7) .However, further experimental and epidemiological research is needed to validate the hand-arm vascular weighting A p and the derived metric A p (8) as preferred measuring methods for the evaluation of daily vibration exposure and the assessment of vibration induced adverse vascular outcomes. Table 1 . Characteristics of the controls and the HTV workers at the cross-sectional survey Data are given as medians (quartiles) or numbers (%).The HTV workers are divided into two sub-groups according to the occurrence of VWF over the study period, (see text for the definitions of HTV, VWF, A h (8), and A p (8)). Table 3 . Relation of cold test outcome at 10°C (%FSBP 10°) to measures of daily vibration exposure expressed in terms of 8-h energy equivalent r.m.s. acceleration calculated by weighting the r.m.s. acceleration magnitudes of the tools according to either the frequency weighting recommended in ISO 5349-1 (A h (8)) or the hand-arm vascular weighting defined in ISO/TR 15870 (A p (8)) Regression coefficients (95% confidence intervals) are estimated by maximum-likelihood random-effects regression models for repeated measures over the follow up period.The likelihood ratio (LR) test for the significance of the measures of daily vibration exposure and the Bayesian information criterion (BIC) for the comparison between models are shown.See text for the definition of %FSBP 10°.a, b Adjusted by survey time and %FSBP 10° at baseline.c Adjusted by age-at-entry, smoking, drinking, body mass index, hand trauma or surgery, systemic disorders, daily use of medicines, leisure activity with vibrating tools, survey time, and %FSBP 10° at baseline. d VWF: symptoms of vibration induced white finger at the medical interview assisted by colour photographs.e p<0.0001 for A f (8) in all models. Table 4 . Relation of cold test outcome at 10°C (DPI 10°) to measures of daily vibration exposure expressed in terms of 8-hr energy equivalent r.m.s.acceleration calculated by weighting the r.m.s.acceleration magnitudes of the tools according to either the frequency weighting recommended in ISO 5349-1 (A h (8)) or the hand-arm vascular weighting defined in ISO/TR 15870 (A p (8)) Regression coefficients (95% confidence intervals) are estimated by maximum-likelihood random-effects regression models for repeated measures over the follow up period.The likelihood ratio (LR) test for the significance of the measures of daily vibration exposure and the Bayesian information criterion (BIC) for the comparison between models are shown.See text for the definition of DPI 10°.a,bAdjusted by survey time and DPI 10° at baseline.cAdjustedbyage-at-entry,smoking, drinking, body mass index, hand trauma or surgery, systemic disorders, daily use of medicines, leisure activity with vibrating tools, survey time, and DPI 10° at baseline.dVWF:symptoms of vibration induced white finger at the medical interview assisted by colour photographs.ep<0.0001for A f(8)in all models.	2023-05-09T06:16:54.607Z	2023-05-08T00:00:00.000	{ "year": 2023, "sha1": "a8ee8ad7fd4c416c1e8bd44a0ce04a57133acda6", "oa_license": "CCBYNCND", "oa_url": "https://www.jstage.jst.go.jp/article/indhealth/advpub/0/advpub_2023-0023/_pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "6d85ae915e229d38c93deb837836e950e9851710", "s2fieldsofstudy": [ "Medicine", "Engineering", "Environmental Science" ], "extfieldsofstudy": [ "Medicine" ] }
239712057	pes2o/s2orc	v3-fos-license	Drought-Hardening Improve Waterlogging Tolerance of Maize at Seedling Stage This study aimed to investigate the stress tolerance of maize by exploring the changes in abscisic acid (ABA) concentration, biomass accumulation, and transpiration rate of maize exposed to drought and waterlogging stress; The experiment was conducted in a controlled greenhouse by setting up a total of 17 treatments of water stress, waterlogging stress, and drought stress followed by waterlogging. A completely randomized block design was employed; Waterlogging limited the formation of maize biomass more than that by water stress. Waterlogging alone (W) inhibited more strongly the growth of the aboveground part than of the underground part, causing a de-crease in the canopy-to-root ratio. However, the canopy-to-root ratio increased under waterlogging after drought. Under drought and waterlogging stress, the ABA concentrations of maize leaves and roots changed gradually, decreasing from the leaves to the root base, then middle root, and nally root tip. Early water stress had a greater effect on leaf ABA concentration than on root ABA concentration, and leaves were the most sensitive to drought stress. Root system was more sensitive to drought stress followed by waterlogging; Moderate controlled drought at the seedling stage can improve their resistance to subsequent waterlogging stress, but the subsequent water-logging stress should not last more than 7 days. Introduction Environmental stresses restrict agricultural development worldwide, which are therefore perpetual topics for discussion. Drought, ood and their secondary disasters are the top issues restricting agricultural development and seriously decrease crop growth and yield [1][2][3][4][5]. Drought stress in early crop growth can promote the emergence of certain metabolites and structural functions in later crop growth, which is bene cial to the improvement of crop stress resistance [6]; for example, it enhances the ability of crop roots to extend into deeper soil layers [7], adjusts the stomatal opening to reduce water loss owing to transpiration [8], and changes the biological properties [9,10]. One important physio-logical response of crops after a drought is the change in the abscisic acid (ABA) concentration, which is considered a characteristic indicator of drought stress in crops. Under drought stress, ABA is synthesized and transported in crop tissues, where it synergistically interacts with water to mediate crop stomatal opening and maintain water balance in crop plants [8,11]. The degree of damage brought by drought to crops depends on the soil water holding capacity, crop type, growth stage, etc. Timely replenishment of water after a drought will enable the crops to recover their growth, whereas excessive water sup-ply after a drought will easily cause waterlogging. Waterlogging weakens the respiration of crop roots and impairs or prevents early stomatal conductance in crops, which is therefore not conducive to stress hardening of crops [11]. Under controlled conditions, crops can receive appropriate amounts of water supply in a timely manner after a drought. However, under the current frequent global extreme climate events, spring drought is often accompanied by rainy weather. The seedling stage of maize coincides with spring droughts and rainy weather, and the seedlings are thus exposed to damaging effects of rainwater logging. Humans are still unable to eliminate the risks of rainwater logging in a timely manner, i.e., they are often able to quickly eliminate waterlogging, but it takes some time for eld waterlogging to drop to safe water levels. Therefore, sustainable and e cient agricultural production in dry areas urgently requires a solution to keep and moderate eld drought and waterlogging under real-time control in extreme weather to stimulate the stress resistance of crops and reduce the damage to their growth. This study aimed to investigate the tolerance of maize seedlings to drought and waterlogging stress. To that end, maize seedlings grown in a greenhouse under constant temperature were exposed to drought and waterlogging stress to determine ABA concentrations and physiological responses. Biomass changes Leaf area is directly related to crop respiration, photosynthesis, and biomass. Too large or too small leaf area will affect the improvement of crop yields. This study showed that for maize at the seedling stage, 65%-Wi and 75%-Wi treatments had no signi cant effect on leaf area. In contrast, waterlogging after severe water stress (55%) affected leaf area development, and waterlogging alone (W) signi cantly inhibited leaf extension ( Figure 1). The most direct effect of stress on plants is the inhibition of crop growth and biomass reduction. In this study, the effects of waterlogging after drought on maize biomass accumulation at the seedling stage varied. The leaf dry matter in different treatments after one day of post-drought waterlogging were ranked as follows: 65%-W1 > 55%-W1 > 75%-W1 > W. After three days of post-drought waterlogging, the leaf mass in treatment 55%-W3 quickly increased. After ve days of post-drought waterlogging, the leaf mass in 65%-W5 was close to the control level. Treatment W showed a signi cant inhibitory effect on leaf mass accumulation (Figure 2 (a)). Waterlogging-alone (W) treatment showed a signi cant inhibitory effect on stem and root biomass accumulation, whereas post-drought waterlogging had no signi cant effect on stem biomass accumulation (Figure 2 (b)). Different degrees of drought and waterlogging stress had a greater impact on the root system when compared with that of stem and leaves of maize at the seedling stage. In Phase I, root biomass increased with water stress: the biomass of the root system under water stress was higher than that under waterlogging-alone (W) stress. During the post-drought waterlogging stress in Phase II, the root biomass accumulation rate in treatment 65%-Wi increased signi cantly, the highest root biomass was measured in 65%-W10 treatment, followed by that in 55%-W10. However, the root biomass under waterlogging-alone (W) stress and higher water content (75%) was relatively small, showing that waterlogging alone or high-water content was not conducive to root development, drought treatments followed by waterlogging inhibited the root biomass accumulation (Figure 2 (c)). The canopy-to-root ratios in different treatments at the end of Phase I were reduced gradually with reducing soil water content from 75% to 55% eld capacity. After 10 days of waterlogging in Phase II, the canopy-to-root ratio was the lowest in waterlogging-alone (W) treatment and the highest in 75%-W10, whereas those in 65%-W10 and 55%-W10 treatments did not signi cantly differ from the control (Figure 2 (d)). Changes in ABA concentration Soil water stress can change the physiological properties of crops and thereby in-crease the ABA concentration in xylem sap of crops. The change in ABA concentrations can be adjusted or corrected via different irrigation methods to optimize the water use e ciency of crops [12,13]. Under the stress of drought followed by waterlogging, the ABA concentrations of maize leaves and roots decreased gradually from the leaves to the root base, middle root, and root tip ( Figure 3). The greater the water stress, the higher the foliar ABA concentration and the larger the difference between the treatments (P<0.05). ABA concentration of the root system did not signi cantly differ between the treatments in Phase I (P<0.05), but it was more affected by water stress than by waterlogging. In Phase II, different durations of waterlogging had different effects on the ABA concentrations of maize. The plant ABA concentration in 75%-Wi treatment gradually decreased with pro-longed waterlogging. The leaf and root ABA concentrations under 75%-W1 treatments were signi cantly higher than those after 3 to 7 days of waterlogging. There was no signi cant difference in leaf and root ABA concentrations after 3, 5, and 7 days of waterlogging. From the rst to fth day of waterlogging, the plant ABA concentrations in 55%-Wi and 65%-Wi treatments decreased with prolonging the waterlogging. During the same period, the ABA concentration of the plant segment from leaves to the root base signi cantly varied with different durations of waterlogging, whereas that of the segment from the middle root to the root tip did not signi cantly vary with different waterlogging periods. The seventh day of waterlogging resulted in the lowest leaf ABA concentrations but the highest root ABA concentration in 55%-W7 and 65%-W7 during the waterlogging period, especially, the ABA concentration increased signi cantly in the segment from the middle root to the root tip in 65%-W7. The leaf ABA concentration in the waterlogging-alone (W) treatment decreased as the duration of waterlogging increased from 1 to 5 days but re-bounded to the highest value on day 7 of waterlogging. In the waterlogging-alone environment, the leaves responded after 7 days of waterlogging stress, whereas the root ABA concentration did not respond to waterlogging stress. It can be observed from the gradients of ABA concentrations that the leaf ABA con-centration increased with the increase in early water stress and so did the ABA concentration of the plant segment from the leaves to roots. The ABA concentration of the leaves was more sensitive to the water environment than that of the roots. Figure 4 shows changes in leaf stomata and plant transpiration under drought and waterlogging stress. Stomatal conductance of leaves under moderate and low water control (65% and 55%) and waterloggingalone (W) stress was inhibited in Phase I. Stomatal conductance under treatments of waterlogging after drought stress increased on the rst to third day of Phase II, and then showed a downward uctuating trend. In 65%-Wi treatment, the stomatal conductance uctuated the most from days 3 to 7 of waterlogging. On the fourth to fth day of waterlogging, the stomatal conductance gradually decreased, whereas the transpiration gradually reached a peak. The stomatal conductance in 75%-Wi treatments uctuated greatly within a short period and lagged behind the stomatal conductance observed in 65%-Wi treatment. The stomatal conductance in other treatments uctuated slightly (Figure 4 (a)). Discussion Drought and waterlogging are frequent natural disasters in spring and summer. The most direct effect of stress on crops is to inhibit crop growth and reduce biomass [14,15]. Therefore, biomass is often used as an important measure of crop resistance to stress. In this study, watering at 65% eld capacity had the most prominent advantage in terms of the leaf, stem, and root dry matter. Waterlogging alone (W) showed a signi cant inhibitory effect on the biomass of maize stems, leaves, and roots, indicating that maize at the seedling stage was more resistant to waterlogging stress after one week of water stress hardening (Fig. 1). Therefore, early moderate drought helps to alleviate or improve the effect of subsequent waterlogging on crop biomass, thereby improving crop resistance to subsequent waterlogging. This, in turn, lays a good foundation for promoting crop yield and ensuring economic bene ts [16][17][18]. The canopy-to-root ratio can re ect the sensitivity of crops to stress. The ratio depends on the duration of the stress, and it is considered an adaptation mechanism to stress [19]. Research has revealed that some crop growth characteristics recovered to the control levels in 7 days after a continuous drought followed by waterlogging [20,21], and another study showed that ABA synthesis induced by water stress is related to the canopy-to-root ratio [19]. Therefore, the canopy-to-root ratio is another important measure in studies of crop response to environmental stress. The results of the present study showed that in Phase I, greater water stress had a greater effect on the growth of shoots and leaves above ground and generated a more developed root system; on the seventh day of waterlogging after drought (65%-Wi and 55%-Wi) in Phase II, the canopy-to-root ratio was close to the control level ( Fig. 1 (d)). The canopy-to-root ratio re ected the physiological regulation effect of crops under stress of drought followed by waterlogging. The main hazard of waterlogging is the inhibition of root function. The early water stress promoted the growth of the root system and expanded the extent of the root functional zone, which laid a good physiological foundation for the hardening of crop vi-ability under later waterlogging stress and had a buffering and regulating effect on later maintenance of waterlogging damage to the root zone. Increasing dry matter accumulation is the main way to improve the crop harvesting potential [22]. In contrast, soil moisture regulation is an important means to interfere with crop product formation. Excessive irrigation or waterlogging stress will cause obvious changes in crop morphology, which is the most intuitively characterized by the deterioration of leaf photosynthesis and corresponding reduction of transpiration [14]. The present study found that maize transpiration was related to the level of drought and waterlogging. The transpiration of maize un-der drought stress followed by waterlogging was higher than that under waterlogging alone (W). When there was no signi cant change in stomatal conductance, the daily plant transpiration decreased by 28.41% on the fth day and by 30.13% on the seventh day of the waterlogging-alone (W) treatment at the seedling stage of maize. The small leaf area under waterlogging alone resulted in a small number of stomata and thereby low transpiration rate. In 65%-Wi, the stomatal conductance gradually decreased on the fourth to fth day of waterlogging, while at the same time the transpiration gradually reached a peak. The difference in plant transpiration among different treatments under drought stress followed by waterlogging become insigni cant starting from day 7 in Phase II, re ecting the hardening of maize to water stress. The physiological stress resistance function manifested itself within 7 days of waterlogging after a drought. As a stress hormone, ABA plays a very important role in regulating the growth and development of plants, especially in many physiological processes of abiotic stress response (such as high salt, low temperature, and drought) [20]. Studies suggest that a variety of abiotic stresses can cause increased ABA concentrations in plants [19,23]. Under water stress, the root system is the rst part to sense changes in soil moisture. The plant ABA concentration originates from the root system, and its change is, therefore, stimulated by soil water stress [13]. Therefore, the change in plant ABA concentration is usually regulated or corrected through different irrigation methods to optimize crop water use e ciency [12]. In this study, water control followed by waterlogging led to a gradual change in the ABA concentration of maize leaves and roots. Speci cally, the ABA concentrations decreased from leaves to the root base, then middle root, and nally root tip. The root system is the source of drought stress signals. Such signals accumulate and spread to the leaves along with the sap ow. Therefore, the leaf becomes the end point of ABA signal accumulation and witnesses the highest ABA concentration. The root system had higher ABA concentrations in 65%-Wi and 55%-Wi than under waterlogging alone (W) on day 7 of waterlogging in Phase II, indicating that drought followed by waterlogging elicits root production of stress signal and changes the physiological properties of the root system. Early water stress awakened the physiological response of maize to the subsequent waterlogging stress. Drought stress had a mediating effect on the physiological functions of maize, and this effect manifested itself in the subsequent stress. Studies on other plants have found that stomatal conductance and transpiration rate of plants are slightly or signi cantly reduced by drought stress and recover to varying degrees within 15 days of re-hydration [9,24]. Drought stress inhibits the increase in leaf area. After rehydration starts, the leaf area shows a short and rapid increase, and the dry matter accumulation rate in-creases, resulting in compensatory growth effects [9,[25][26][27][28]. In the present study, the ABA concentration of leaves under water stress (55% and 65% eld capacity) was signi cantly higher than that under non-stress treatments (75% eld capacity). With the increase in water stress, the ABA concentration of leaves increased, the leaf stomatal conductance and plant transpiration decreased, but the inhibitory effect of water stress on leaf area and dry matter was not signi cant. On day 1 to 5 of waterlogging, the ABA concentration in the stem between the leaves and the root system increased with increasing water stress. In this period, the leaf stomatal conductance, plant transpiration, and dry matter under 65%-Wi treatment was also recovered and increased, whereas the recoverability of these indexes in 55%-Wi treatment was relatively weak. On day 7 of waterlogging, while the leaf ABA con-centration continued to decrease, the physiological function of the root system was induced and stimulated, resulting in increased root ABA concentration. Consequently, the ABA concentration gradient between leaves and roots decreased, and the ABA concentrations of leaves and roots were relatively balanced. The leaf stomatal conductance and plant transpiration in 65%-W7 and 55% -W7 treatments slowly decreased. The dry matter accumulation rate in 65%-W7 slowed down, and that in 55%-W7 decreased signi cantly. Rehydration effects occurred within 5 days of waterlogging after moderate drought stress (65% eld capacity), and each parameter increased to the compensation point. A more severe water stress (55% eld capacity) reduced the compensation effect, which is not conducive to biomass formation. After 7 days of waterlogging following drought stress, the negative effects of waterlogging stress began to appear. Therefore, waterlogging stress after a drought period at the seedling stage of maize should not exceed 7 days. Experimental design The experiment was conducted in a temperature-controlled greenhouse with PPFD of 400 ~ 600 µmol m − 2 ·s − 1 , 14 h photoperiod, and day and night temperature of 19 ~ 24 ℃. Maize variety sweet corn F1 Early Sunglow, a commercialized seeds acquired from the lab in Environment Centre, Lancaster University, was planted in black cylinders made of PVC plastic, with an outer diameter of 6.6 cm, thickness of 2 mm, and height of 24.5 cm. The bottom of each cylinder was sealed with a stainless-steel net to facilitate drainage and stabilize the cylinder. The cultivated organic loam John Innes No. 2 was used as experimental soil. This soil type is traditional British soil compost containing loam, sphagnum peat, grit, humus, and fertilizers that promote healthy growth of plants. Owing to its good performance, John Innes No. 2 has been widely used in the UK for more than 60 years. Each cylinder was lled with 1050 g of the experimental soil. The soil was lled and compressed in layers so that the soil bulk density in each cylinder was the same. Each cylinder was lled with soil to about 2.5 cm below its edge. Two maize seeds were sown in each cylinder and watered with 50 mL of water. The water amount was determined with a measuring cup. During the waterlogging test, the soil in each pot was covered with a plastic to prevent water from seeping out. Soil water was controlled from the three-leaf stage of maize, and the experiment was divided into two phases, i.e., Phase I (7 days) and Phase II (10 days). In Phase I, the soil water content was kept at 55%, 65%, and 75% of the eld capacity; in the waterlogging-alone (W) treatment, a water layer of 1 cm was maintained on soil surface. In Phase II, the treatments from Phase I were subjected to waterlogging for 1, 3, 5, 7, and 10 days (55%-Wi, 65%-Wi, and 75%-Wi, where i is the number of days a plant was subjected to waterlogging; Table 1). The treatment of water control at 75% eld capacity (optimal soil water content for growth of maize) was used as a control. There were 17 treatments in total, each repeated three times. Methods During the experiment, the top of each cylinder was sealed with tin foil to expose only the maize plants. The total mass of the cylinders was weighed every morning from 8:00 to 9:00 to calculate the daily water loss owing to plant transpiration. The lost water was then supplemented. Maize was harvested at the end of the experiment. Aboveground parts of the seedlings were cut and separated into stem and leaves in each pot. The soil from each pot with the root zone was divided into four segments: 0 ~ 5 cm, 5 ~ 10 cm, 10 ~ 15 cm, and 15 ~ 22 cm from the soil surface. The roots in each segment were then washed. Leaf area was measured by an automatic leaf area meter. Stem, leaf, and root samples were dried at 65 ℃ for 72 h, and their dry matter was then determined. ABA concentration in leaves and roots was measured by radioimmunoassay using the monoclonal antibody AFRC MAC 252 [29]. The youngest and fully expanded lea et was harvested for ABA measurement. The sampling was conducted at the same time (10:00 ~ 10:30) on each harvesting day to avoid diurnal effects on foliar ABA concentration. Lea ets and roots (of the same plant) were sampled, frozen in liquid nitrogen, freeze-dried for 48 h, and then nely ground. Small amounts of samples (10 ~ 15 mg dry weight for lea ets, and 30 ~ 40 mg dry weight for roots) were used for ABA analysis. The powdered samples were diluted with deionized, distilled water (1:70 for lea ets, and 1:25 for roots) and placed on a shaker in a cold room (4°C) overnight to extract ABA. A standard curve was determined with standards in a serial dilution of synthetic unlabeled (±)-cis, trans-ABA (Sigma Let., Dorset, UK). ABA concentration was calculated by reference to the standard curve after linearization using the 'logit' transformation. Statistical analysis All data were analyzed by one-way analysis of variance (ANOVA). Less signi cant difference (LSD) and Student's t-test were carried out (P < 0.05) using Statistica 10.0 (StatSoft, Inc., Tulsa, OK, USA). Conclusions At the seedling stage of maize, early moderate water stress (65% eld capacity) pro-moted leaf biomass accumulation under subsequent waterlogging stress. The canopy-to-root ratio under drought stress followed by waterlogging was higher than that under waterlogging alone (W). The solid matter foundation improved the stress resistance of maize under drought stress followed by waterlogging. Under drought and waterlogging stress, the ABA concentration of maize leaves and roots gradually decreased from leaves to the root base, then middle root, and nally root tip. Maize leaves were the most sensitive to drought stress, and the root system had a certain response to post-drought waterlogging. At the seedling stage, moderate drought stress on maize had a mediating effect on its physiological functions, and this effect manifested itself under later stress. Physiological rehydration effects occurred in maize crops within 5 days of waterlogging after moderate drought (65% eld capacity), and the biomass witnessed a compensatory increase. A more severe stress (55% eld capacity) reduced the compensatory effect. Starting from day 7 of waterlogging that was conducted after water stress, damage to maize crop owing to waterlogging stress began emerge. Therefore, the waterlogging stress should not exceed 7 days after a drought at the seedling stage of maize. Data Availability Statement: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.	2021-09-25T15:53:58.793Z	2021-08-25T00:00:00.000	{ "year": 2021, "sha1": "0545a215d3fa38b158a3a3dee234a23796e39d89", "oa_license": "CCBY", "oa_url": "https://www.researchsquare.com/article/rs-813730/v1.pdf?c=1636964060000", "oa_status": "GREEN", "pdf_src": "ScienceParsePlus", "pdf_hash": "92e6ba4d7c51a417d6c56ad851b63d2ee5185cb5", "s2fieldsofstudy": [ "Agricultural and Food Sciences" ], "extfieldsofstudy": [ "Biology" ] }
1199173	pes2o/s2orc	v3-fos-license	Synthesis and Study of the Effect of Ba2+ Cations Substitution with Sr2+ Cations on Structural and Optical Properties of Ba2−xSrxZnWO6 Double Perovskite Oxides (x = 0.00, 0.25, 0.50, 0.75, 1.00) The effect of Sr2+ substitution on the morphology, crystal structure, and optical properties of double perovskite oxide Ba2−xSrxZnWO6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) were investigated. Scanning electronic microscopy demonstrated that all samples have similar microstructure morphology but differ in the range of grain sizes. X-ray diffraction measurements indicated that these materials crystallize in a (Fm-3m) cubic crystal structure, and also confirmed the tolerance factor. Rietveld analysis revealed that the lattice parameter decreased from 8.11834 to 8.039361 Å when the substitution of Ba2+ with Sr2+ cations increased from zero to 100%. Fourier transform infrared (FTIR) and Raman spectroscopies displayed a symmetric stretching vibration of WO6 octahedra at 825 cm−1, and an anti-symmetric stretching mode of WO6 was observed by FTIR at 620 cm−1. A strong peak at 420 cm−1 was also observed in the Raman spectra and is due to the W–O–W bending vibration modes. UV-Vis diffuse reflectance spectroscopy was carried out for the series, and the band gap energy decreased from 3.27 eV for Ba2ZnWO6 to 3.02 and 3.06 eV for Ba1.75Sr0.25ZnWO6 and Ba1.5Sr0.5ZnWO6, respectively. The excitation and emission photoluminescence properties were investigated at room temperature. A variety of devices and methods have previously been used to find and characterize new double perovskite compounds at high temperatures. Zaraq et al. [1] synthesized SrCaCoTeO 6 and SrCaNiTeO 6 compounds and used X-ray diffraction (XRD) and scanning electron microscopy (SEM) to describe their crystal structure and phase transition. Chufeng Lau et al. [2] used XRD, photoelectron spectroscopy, Scanning Electron Microscopy The ESM images of the Ba 2−x Sr x ZnWO 6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) series are displayed in Figure 1a-e. The morphologies of all samples are identical and they appear to be highly homogeneous with no impurities. It is observed that, in all samples, the size of the particles is large and they are aggregated in groups, which is due to the higher preparation temperature. Chunfeng Lan et al. [2] observed an equivalent effect of temperature in the morphology of La 2 NiMnO 6 double perovskite oxide. Furthermore, the images reveal the presence of fine fragments that are produced during the preparation grinding. Each of the samples has grains of various sizes, i.e., Ba 2 ZnWO 6 has 1-3 µm grains, Ba 1.75 Sr 0.25 ZnWO 6 has 1-5 µm grains, Ba 1.5 Sr 0.5 ZnWO 6 has 1.5-4 µm grains, Ba 1.25 Sr 0.75 ZnWO 6 has 2-8 µm grains, and BaSrZnWO 6 has 2-7 µm grains. An Energy-dispersive X-ray spectroscopy (EDX) analysis is conducted with each sample using the SEM images. Figure 1e presents the energy dispersive X-ray spectrum from the element that formed the BaSrZnWO 6 sample. All EDX graphs confirm that all samples contain elements of the raw material preparation composition and a proportion approximating the input quantities to configure each sample with a small error, which refers to the homogeneity and crystal purity. X-ray Powder Diffraction The X-ray diffraction data of the perovskite oxide compounds are essential in determining the crystalline structure of the samples in terms of the Bravais lattice, atomic position, lattice parameter, and space group. Many studies refer to the importance of the study of material structure since they govern the other properties of the materials [17]. The XRD of the Ba2−xSrxZnWO6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) double perovskite oxide series prepared by the solid-state reaction is shown in Figure 2. The BaWO4 and Ba2WO5 phases displayed as minor peaks at low intensity in the XRD pattern shown in Figure 1 are attributed to the impurities in the Ba2−xSrxZnWO6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) structure around 26° and 28° [18]. A plus sign and a star are used to depict them in Figure 1 when 2 is around 26° and 28° for BaWO4 and Ba2WO5, respectively. The XRD data of each sample in the series are refined by the Rietveld method using the FullProf program. Table 1 displays the atom X-ray Powder Diffraction The X-ray diffraction data of the perovskite oxide compounds are essential in determining the crystalline structure of the samples in terms of the Bravais lattice, atomic position, lattice parameter, and space group. Many studies refer to the importance of the study of material structure since they govern the other properties of the materials [17]. The XRD of the Ba 2−x Sr x ZnWO 6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) double perovskite oxide series prepared by the solid-state reaction is shown in Figure 2. The BaWO 4 and Ba 2 WO 5 phases displayed as minor peaks at low intensity in the XRD pattern shown in Figure 1 are attributed to the impurities in the Ba 2−x Sr x ZnWO 6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) structure around 26 • and 28 • [18]. A plus sign and a star are used to depict them in Figure 1 when 2θ is around 26 • and 28 • for BaWO 4 and Ba 2 WO 5 , respectively. The XRD data of each sample in the series are refined by the Rietveld method using the FullProf program. Table 1 Figure 3 shows the XRD refinement of BaSrZnWO 6 , which is represented by a (Fm-3m) cubic structure with lattice parameters a = b = c = 8.039361 Å and α = β = γ = 90 • . Identical results are obtained for Ba 2 ZnWO 6 using single-crystal X-ray and neutron diffraction [19]. Furthermore, the Ba 2−x Sr x MgTeO 6 series was found in a (Fm-3m) cubic crystal structure [20]. The crystallite size was calculated from the Full width at half maximum (FWHMs) at the major peaks at (220) for the Ba 2−x Sr x ZnWO 6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) double perovskite series using the Scherer equation [21], which was observed to vary between 47.41 and 105.9 nm for the samples. where D is the crystalline size, λ is the wavelength (1.5405 Å), β is the full width at half maximum, and θ is the diffraction angle. The tolerance factor was found to be between 0.998 and 1.007, which can be calculated by coordinates of all the samples obtained in the (Fm-3m) cubic crystal structure. Figure 3 shows the XRD refinement of BaSrZnWO6, which is represented by a (Fm-3m) cubic structure with lattice parameters a = b = c = 8.039361 Å and α = β = γ = 90°. Identical results are obtained for Ba2ZnWO6 using single-crystal X-ray and neutron diffraction [19]. Furthermore, the Ba2−xSrxMgTeO6 series was found in a (Fm-3m) cubic crystal structure [20]. The crystallite size was calculated from the Full width at half maximum (FWHMs) at the major peaks at (220) for the Ba2−xSrxZnWO6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) double perovskite series using the Scherer equation [21], which was observed to vary between 47.41 and 105.9 nm for the samples. where D is the crystalline size, λ is the wavelength (1.5405 Å), β is the full width at half maximum, and θ is the diffraction angle. The tolerance factor was found to be between 0.998 and 1.007, which can be calculated by Table 2 displays the tolerance factor and parameter of the crystal structure of Ba2−xSrxZnWO6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) using the Rietveld method of refinement. The unit cell volume decreases with an increasing substitution as a result of the larger ionic radius of the Ba 2+ cation than that of Sr 2+ . Table 2 displays the tolerance factor and parameter of the crystal structure of Ba 2−x Sr x ZnWO 6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) using the Rietveld method of refinement. The unit cell volume decreases with an increasing substitution as a result of the larger ionic radius of the Ba 2+ cation than that of Sr 2+ . The FTIR spectra identify the crystal structure of the perovskite structure materials that have characteristic absorption bands in the 850-400 cm −1 region [22]. The strong high-energy anti-symmetric stretching mode of the WO 6 octahedral displayed at 620 cm −1 is due to the higher charge of the tungsten cations. The symmetric stretching vibration of the WO 6 octahedral appears as a high-intensity band at about 825 cm −1 . Figure 4 shows the transmittance of the Ba 2−x Sr x ZnWO 6 double perovskite series versus wave number, and all the samples confirm the molecular bands on the perovskite oxide structure [17]. characteristic absorption bands in the 850-400 cm −1 region [22]. The strong high-energy anti-symmetric stretching mode of the WO6 octahedral displayed at 620 cm −1 is due to the higher charge of the tungsten cations. The symmetric stretching vibration of the WO6 octahedral appears as a high-intensity band at about 825 cm −1 . Figure 4 shows the transmittance of the Ba2−xSrxZnWO6 double perovskite series versus wave number, and all the samples confirm the molecular bands on the perovskite oxide structure [17]. Raman Spectroscopy The Raman spectra of the samples are shown in Figure 5 for Ba2−xSrxZnWO2 (x = 0.00, 0.25, 0.50, 0.75, 1.00). For all samples, the Raman modes are classified into two types of lattice vibration-the W-O-W bending vibration in the 200-500 cm −1 region and the W-O stretching mode between 700 and 950 cm −1 . This result was observed in many studies on double perovskite oxides [22,23]. Ba has a larger ionic radius (149 Å) than that of Sr (132 Å). When the Ba substitution increases, the effect of the Ba radius is reflected by a decrease in both the bending and stretching modes of the W-O bonds. A redshift in the Raman energy is also observed. Raman Spectroscopy The Raman spectra of the samples are shown in Figure 5 for Ba 2−x Sr x ZnWO 2 (x = 0.00, 0.25, 0.50, 0.75, 1.00). For all samples, the Raman modes are classified into two types of lattice vibration-the W-O-W bending vibration in the 200-500 cm −1 region and the W-O stretching mode between 700 and 950 cm −1 . This result was observed in many studies on double perovskite oxides [22,23]. Ba has a larger ionic radius (149 Å) than that of Sr (132 Å). When the Ba substitution increases, the effect of the Ba radius is reflected by a decrease in both the bending and stretching modes of the W-O bonds. A redshift in the Raman energy is also observed. Figure 6 presents the diffuse reflectance spectrum of the Ba 2−x Sr x ZnWO 6 series at room temperature in the 200-800 nm range. The strong absorption band observed at 300-450 nm refers to the absorption edge in tungsten due to the charge transfer transition of W 6+ − O 2− in the lattice from the highest filled molecular orbital 2p of oxygen to the lowest empty molecular orbital 5d of tungsten [14,19,24]. The absorption band has a blue shift with an increasing substitution ratio of Ba 2+ with Sr 2+ cations. Figure 6 presents the diffuse reflectance spectrum of the Ba2−xSrxZnWO6 series at room temperature in the 200-800 nm range. The strong absorption band observed at 300-450 nm refers to the absorption edge in tungsten due to the charge transfer transition of − in the lattice from the highest filled molecular orbital 2p of oxygen to the lowest empty molecular orbital 5d of tungsten [14,19,24]. The absorption band has a blue shift with an increasing substitution ratio of Ba 2+ with Sr 2+ cations. The absorption coefficient can be calculated for the Ba 2−x Sr x ZnWO 6 series from the diffuse reflectance data using the Kubelka-Munk function [25]: UV-Visible Diffuse Reflectance Spectroscopy where F(R ∞ ) is the KM function, α is the absorption coefficient, s is the scattering coefficient, and R is the reflection coefficient. The absorbance (f (R)hυ) in relation to wavelength is shown in Figure 7. [19] according to the relationship E g = 1240/λ (λ is the absorption edge wavelength and E g is the band gap [26]). In addition, the band gap energy was calculated for the samples using the Tauc plot [18], as shown in Figure 8, according to Equation (4). where hν is the incident photon energy, A is a proportional constant, E g is the band gap energy, and n takes values of 2 or 0.5 for indirect and direct transitions, respectively. Table 3 illustrates the bang gap energy according to the absorption edge and Tauc plot. The results of UV-vis diffuse reflectance and the optical energy gap of the sample series indicate that they can be classified as semiconductor materials [19,27]. where ℎ is the incident photon energy, A is a proportional constant, is the band gap energy, and n takes values of 2 or 0.5 for indirect and direct transitions, respectively. Table 3 illustrates the bang gap energy according to the absorption edge and Tauc plot. The results of UV-vis diffuse reflectance and the optical energy gap of the sample series indicate that they can be classified as semiconductor materials [19,27]. Initially, the materials were used as purchased from Alfa Acer with a purity of 99.99%. Several different treatments of the samples were conducted to achieve the crystal structure. The raw materials were mixed and ground in an agate mortar with acetone, kept in crucibles, and subsequently heated in air at 800 • C for 12 h twice. The sample pellets were prepared in a round shape and heated at 1000 • C twice, following which the same procedure was repeated at 1200 • C. Between the steps for the heating treatment, the sample was ground for 2 h with acetone to increase the homogeneity of the sample at a rate of 10 • C per minute during the heating and cooling process. Sample Characterization A Jeol JSM-6360 (JEOL Inc. Peabody, MA, USA) and high-resolution Stereo Scan LEO 440 SEM (LEO, Austin, TX, USA) were used to investigate the morphology and determine the homogeneity of the samples, as well as to obtain crystal-scale crystallization. At room temperature, the XRD data were recorded with a Bruker: D8 Advance (Bruker-Axs, Madison, WI, USA) using CuKα radiation (λ = 1.5406 Å) with a nickel filter. At 40 kV and 40 mA, data were collected for 2θ at 0.02-step sizes and 5-s count times in the 20 • -80 • range. The XRD data were analyzed using the Rietveld refinement method with the FullProf Suite program [31]. The crystalline size (D) [32] was calculated using the Debi-Scherer equation for all samples. At room temperature, the transmittance mode was investigated for all samples using the Satellite FTIR 5000 (ARCoptix S.A, Neuchatel, Switzerland) (with a wavelength range of 400-4000 cm −1 ) [33], where the important bands and peaks of the perovskite structure can be assigned. Using FTIR spectroscopy collected using the KBr pellet method, the material was mixed with KBr at 1:100 ratios for the FTIR measurement in the range of 400-2000 cm −1 . The Raman spectra were collected in Raman HR (Stellarnet Inc., Tampa, FL, USA), using the high resolution Raman spectrometer with a range of 200-2200 cm −1 at 785 nm with a resolution of 4 cm −1 . The Raman probe attaches to the laser via FC/APC and the spectrometer via SMA 905, and has integrated Raman filters and optics with a working distance to the sample of 4.5 mm, configured for the 785 nm laser. A UV-Vis spectrophotometer (UV-2550, Shimadzu, Chiyoda-ku, Tokyo) using BaSO 4 as a reference is used to calculate the UV-Vis diffuse reflectance spectrum at room temperature. In addition, the UV-Vis reflectance spectrum is converted to absorbance using the Kubella-Munk method to estimate the edge of absorption and band gap of the Ba 2−x Sr x ZnWO 6 double perovskite powder series. A Perkin Elmer LS55 fluorescence spectrometer (Perkin-Elmer, Wokingham, UK) was used to investigate the emission and excitation of the Ba 2−x Sr x ZnWO 6 double perovskite series at room temperature. Conclusions The Ba 2−x Sr x ZnWO 6 (x = 0.00, 0.25, 0.50, 0.75, 1.00) double perovskite series was prepared by the solid-state reaction technique. In addition, techniques such as X-ray diffraction, scanning electronic microscopy, Fourier transform infrared spectra, Raman spectra, UV-Vis diffuse reflectance, photoluminescence excitation, and emission spectra were investigated. SEM revealed that the prepared Ba 2−x Sr x ZnWO 6 samples crystallized at micrometer scales, where the crystal structure of the samples was determined by XRD as a cubic Fm-3m space group. The lattice parameter decreased with an increase in the proportion of substitution from x = 0 to 1, where the result of FTIR also confirmed the double perovskite structure. The Raman spectra of W-O bonds indicated a systematic decrease and re-shifting with an increasing Ba substitution. Strong UV-Vis absorption was found between 350 and 410 nm and smaller optical bandgap energy, 3.02 eV, was found for Ba 1.75 Sr 0.25 ZnWO 6 compared to the Sr-free sample. The excitation photoluminescence spectra displayed broad bands between 260 and 320 nm, which were assigned to the charge transfer band of Ba 2−x Sr x ZnWO 6 . Ba 2−x Sr x ZnWO 6 displays photoluminescence in the near-UV and visible region. This feature makes Ba 2−x Sr x ZnWO 6 a potential semiconductor in optoelectronics applications.	2017-05-22T10:33:02.151Z	2017-04-28T00:00:00.000	{ "year": 2017, "sha1": "3771f7d2a04259f1c07631777141248f37e0fb73", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1996-1944/10/5/469/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "3771f7d2a04259f1c07631777141248f37e0fb73", "s2fieldsofstudy": [ "Materials Science", "Physics" ], "extfieldsofstudy": [ "Chemistry", "Medicine" ] }
119654639	pes2o/s2orc	v3-fos-license	A general limit lifting theorem for 2-dimensional monad theory We give a definition of weak morphism of $T$-algebras, for a $2$-monad $T$, with respect to an arbitrary family $\Omega$ of $2$-cells of the base $2$-category. By considering particular choices of $\Omega$, we recover the concepts of lax, pseudo and strict morphisms of $T$-algebras. We give a general notion of weak limit, and define what it means for such a limit to be compatible with another family of $2$-cells. These concepts allow us to prove a limit lifting theorem which unifies and generalizes three different previously known results of $2$-dimensional monad theory. Explicitly, by considering the three choices of $\Omega$ above our theorem has as corollaries the lifting of oplax (resp. $\sigma$, which generalizes lax and pseudo, resp. strict) limits to the $2$-categories of lax (resp. pseudo, resp. strict) morphisms of $T$-algebras. Introduction The notion of a 2-monad T in an arbitrary base 2-category K goes back to at least [12], we refer the reader to the expository paper [8, §3] for a detailed account of this theory in its early stages. The relevance of 2-dimensional monad theory, that is the study of algebras of a 2-monad and its morphisms, as a branch of 2-dimensional universal algebra is nowadays very well established. The classic article [3] is a standard reference, and we will use most of the notation as it is there. It is well known that the notion of strict algebra morphism coming from Cat-enriched monad theory has to be relaxed in order to go beyond this theory and into the real and deep 2-dimensional one (see [3], [9] and also [8, 3.5] for examples). In [3, §1.2] lax and pseudomorphisms of T -algebras are introduced. Instead of requiring the commutativity of a diagram, lax morphisms have a structural 2-cell, which is required to be invertible for the pseudomorphism notion (which are referred to simply as morphisms due to its relevance for applications). Their choice of the notation T -Alg for the 2-category of pseudomorphisms of T -algebras is thus appropriate, but the notation T -Alg p is better suited for our point of view in this article, which is to consider all these notions of morphisms as particular cases of a general one. We do this in Section 2, by considering an arbitrary family of 2-cells where the structural 2-cells are required to belong to. The technique of fixing an arbitrary family of 2-cells and considering weak morphisms and (conical) weak limits with respect to it is, as far as we know, due to Gray [6]. A significant fact, observed by Street in [11], is that conical weak limits in this sense are as general as weighted strict limits (i.e. 2-limits, as considered in [7]), in the sense that any such limit can be expressed as one of the other type. Informally speaking, we may say that the weakening of the limit allows to work with only conical ones, without losing generality. We give in Section 3 a notion of weighted weak limit, appropriate for working with weak algebra morphisms, which has both notions above (that is, conical weak limits and weighted strict limits) as particular cases in a direct way. If we interpret the result of [11] in this context (we do this in §3.2), we obtain that every weighted weak limit can be expressed as a conical weak limit. This result allows us to apply our Theorem 5.1, which is a result for conical weak limits, to general weighted ones. We note that, since every weighted weak limit can be expressed as a conical weak limit, using [11] again we have that ultimately it is a strict limit, but it is convenient to disregard this fact and work solely with the weak limit. This is a situation that has previously happened in the literature with lax and pseudolimits: while ultimately being strict limits, these are the appropriate notions for many aspects of 2-dimensional category theory and should not be considered as strict limits for these applications. An important result for monad theory is the lifting of limits along the forgetful functor of the categories of algebras. To this subject we devote Sections 4 and 5. The question of which limits can be lifted has been analysed for various cases, and it is also the subject of the recent article [10], whose introduction we also suggest to the reader. • For the strict theory (or more generally for the V-enriched theory, [5]), it has been long known that all strict limits lift. • For the 2-category of pseudomorphisms, this is treated in [3, §2]. Now it is no longer the case that all strict limits lift. First products, inserters and equifiers are lifted, and then this is used to lift other limits which are constructed from them, their main result being the lifting of pseudolimits, since these are the most important ones for its applications. • Limits in the 2-category of lax morphisms are considered of secondary interest in [3, Remark 2.9], but an interest in them has justifiedly grown since then. In [9] this subject is investigated, its most general theorem being the lifting of oplax limits. The article [10] can be considered as a continuation of the results of [3] and [9], though in a direction quite different to ours. Their main result is a complete characterization (in terms of the weights) of those 2-limits which lift to the 2-category T -Alg ℓ of lax morphisms of algebras for any 2-monad T . They do so by introducing a particular base category F, and working in F-enriched category theory, in particular with F-limits. This allows them to express in the language of enriched category theory properties about the distinguished 2-subcategory of T -Alg ℓ consisting of the strict algebra morphisms. A motivation for these concepts lies in the lifting results of [9, §4], where the hypothesis for the lifting of inserters and equalizers to T -Alg ℓ involve strict morphisms of algebras, these results are generalized to the F-enriched case in [10, 3.5.3,3.5.4]. The main reason for the present paper's approach to this subject lies in my motivation, which was to obtain a proof, as simple as possible, of the lifting of σ-limits (see [4]) to the T -Alg p 2-categories, and of the properties of such lifted limits. This result interested me particularly, as it allows for a different proof of [4,Coro. 2.6.4], which is a fundamental result for the theory of flat 2-functors that we develop there. Though, with a careful checking, one could deduce this result from the results of [3], I felt that this didn't lead to a satisfying answer to the question of why such limits can be lifted. There is an intuitive order between the strictness of the different 2-categories of morphisms of algebras, and I began thinking (inspired in part by our previous work that led to [4,Prop. 2.6.2]) if one could think of such an order for the different types of limits. This is what led me to consider the general notions of ω-T -morphism and σ-ω-limit, which allowed me to formalize the intuition that, the more we relax the notion of morphism, the less limits we can lift. Corollary 5.5 expresses this idea formally and clearly (relaxing the morphisms corresponds simply to augmenting the family Ω ′ of 2-cells), and while it is not the most general result of the paper, it was the most satisfactory for me. In particular, it provided me with an answer to my question (see Remark 5.12): σ-limits can be lifted to T -Alg p because they are σ-ω-limits with respect to Ω p . One fact that distinguishes the lax theory from the other two, as shown in the three items above, is the appearance of the "op" prefix. One cannot avoid to reverse the direction of some 2-cell when working with lax morphisms; if instead of oplax one wishes to lift lax limits, then one must consider colax instead of lax morphisms ([9, Theorem 4.10]). In fact, it is convenient to think that the "op" is always present, but can be ommitted when restricting to morphisms with invertible structural 2-cells (that is, pseudomorphisms and strict morphisms). As it is usually the case, the study of a more general theory yields light on some aspects of its particular cases as well; here in the proof of Theorem 5.1 regarding the lifting of weak op-limits, the necessity of reversing a direction is made transparent, see (5.2). Theorem 5.1 is our main contribution in this article, and it has the three main results of the items above as corollaries. Furthermore, unlike the one in [3, Theorem 2.6], its proof is direct, in the sense that it does not depend on the construction of the limit to be lifted in terms of other limits. Also, Propositions 4.2, 4.3 and 4.4 generalize the corresponding results of [3] and [9] regarding the lifting of strict limits. A comment regarding its relevance can be found at the beginning of Section 4. A fundamental notion which allows these generalizations is that of Ω-compatibility (with respect to a family Ω of 2-cells) for a limit, see §3. 1. Informally, what it means for a limit to be Ω-compatible is that the bijection at the level of 2-cells given by its 2-dimensional universal property restricts to Ω. This notion holds automatically if Ω consists of the invertible 2-cells, and this fact is implicitly used in the proofs of the first three propositions of [3, §2]. By making it explicit, its generalizations in Section 4 follow, with proofs that involve the same computations as the ones in [3]. We note that the proofs of the results in [9] that we generalize here are substantially different to those of [3], and thus of ours. Also, our results have as corollaries the results of [9] in the stronger form in which they appear in [9, §6], requiring the algebra morphisms to be pseudo, not necessarily strict. Regarding [10] again, since that article restricts to the case of enhanced 2-categories of strict and lax morphisms (though the authors hope to address other cases in future papers), I believe thus our results of Section 4 probably admit analogous F-enriched versions, and this should be relevant for the case of enhanced 2-categories of pseudo and lax morphisms. There is an extra property, present in the mentioned results of [3] and [9], of the projections of the lifted limits. Namely, they are strict, and they (jointly) detect strictness. We also generalize this property under the light of Ω-compatibility: we show as part of our results that the projections of the lifted limits are strict, and that they (jointly) detect Ω-ness if the limit in the base category is Ω-compatible. Since for the strict case (that is, when Ω consists only of the identities) every limit is Ω-compatible, ours is indeed a generalization. Weak morphisms of algebras We fix an arbitrary family Ω of 2-cells of a 2-category K, closed under horizontal and vertical composition, and containing all the identity 2-cells. Whenever we consider families of 2-cells, we will assume these axioms to hold. We refer to [3] for the basic definitions of 2-dimensional monad theory. We consider a 2-monad T = (T, m, i) on K, and strict algebras of T . In this section we will define the 2category of weak morphisms of T -algebras (with respect to Ω), which has as particular cases the 2-categories of strict, pseudo and lax morphisms of T -algebras. 2-cells between ω-morphisms (f, f ) α =⇒ (g, g) are arbitrary (that is, not necessarily in Ω) 2-cells f α =⇒ g required to satisfy the equation In this way, with the evident laws of composition, we have a 2-category T -Alg Ω ω of T -algebras and ω-morphisms, and a forgetful 2-functor T -Alg Ω ω U Ω ω − − → K. We refer to these 2-categories as 2-categories of weak T -algebra morphisms, or ω-T -morphisms. Whenever possible, we will omit Ω from the notation and write T -Alg ω , U ω . Remark 2.3. We note (see also [3, p.3], [9, Remark 2.1]) that reversing the direction of f yields a notion of coω-morphism (f, f ), we denote the 2-category so defined by T -Alg coω . Since a coω-morphism (f, f ) is just a ω-morphism for the 2-monad T co in K co , then we have T -Alg co coω = T co -Alg ω . Note that if Ω has only invertible 2-cells, then taking the inverse of f yields an isomorphism of categories T -Alg coω ∼ = T -Alg ω . Example 2.4. By considering the families Ω γ , γ = s, p, ℓ consisting respectively of the identities, the invertible 2-cells and all the 2-cells of K, we recover the 2-categories T -Alg γ introduced in [3] (note that their T -Alg is our T -Alg p ), that is: In this way, all the known examples of such 2-categories, found for example in [3, §6], [9, §1, §5], are 2-categories of ω-T -morphisms. σ-ω-limits We fix throughout this section a family Σ of arrows of a 2-category A, closed under composition and containing all the identities, and a family Ω of 2-cells of a 2-category B. In this section we will define the notion of weighted σ-ω-limit (with respect to Σ and Ω), which is a generalization of weighted strict limits ( [7]) and of Gray's cartesian quasi-limits ( [6]). The notion of weighted strict limit can be recovered by considering a particular choice of Σ and Ω, and the notion of cartesian quasi-limit is exactly the notion of conical σ-ω-limit, thus it corresponds to the particular weight △1 constant at the unit category. On the other hand, we show in Proposition 3.18 that any weighted σ-ω-limit can be expressed as a conical σ-ω-limit (thus as a quasi-limit), and therefore by a result of Street ([11, Theorem 14]) also as a weighted strict limit. Though ultimately we are working only with strict limits, the notion of (weighted) σ-ω-limit turned out to be the most convenient one for working with 2-categories of weak morphisms. We begin by recalling from [6] (while adapting the notation to one more adapted to the current literature, in particular to [4]) the concepts of σ-ω-natural transformation and conical σ-ω-limit (in both cases with respect to Σ and Ω). The case in which Ω = Ω p consists of the invertible 2-cells is considered with great detail in [4], where these concepts are referred to as σ-natural transformation and conical σ-limit. -natural transformations is the same as a modification between the underlying lax natural transformations, that is a family {θ has as objects the 2-functors, as arrows the σ-ω-natural transformations and as 2-cells the modifications. Whenever possible, we will omit Σ and Ω from the notation. We refer to these 2-categories as 2-categories of σ-ω-natural transformations. Note that s is the bottom element of L A,B , and ℓ is the top one. For any label ε = (σ Σ , ω Ω ) ∈ L A,B we denote Hom ε (A, B) = Hom Σ,Ω σ,ω (A, B). In this way, for each choice of γ as above, we recover the usual 2-categories Hom γ (A, B) as 2-categories of σ-ω-natural transformations. in the sense that for each B ∈ B, post-composition with π is an isomorphism of categories We refer to the arrows π A , for A ∈ A, as the projections of the limit. We should warn the reader that in [4] we chose to use a letter w and denote w-σ-cone to indicate that the cone is taken with respect to a weight, in order to distinguish it from a (conical) σ-cone. For obvious reasons, it is convenient not to use such a notation here. Thus, in the present article, σ-ω-cones can be either conical or weighted, with the presence of a weight indicating the latter. We will now generalize this notion to the notion of weighted σ-ω-limit. We note that the category of cones that is needed in order to define weighted σ-ω-limits can't be given (at least in an evident way) as a Hom σ,ω category, as it is done for strict weighted limits and for σ-limits (in particular for weighted pseudo and lax limits as in [7, §5]), since we do not have a family of 2-cells of Cat to play the role of Ω. What we do instead is to describe with detail the notion of lax cone (with respect to a weight), and this allows us to define which of the structural 2-cells should be in Ω for it to be a σ-ω-cone. We remark that we could consider only the conical case and use (3.19) in order to apply this definition to the weighted case, but we prefer the more explicit approach of Definition 3.5 (and also of Definition 3.14 for the notion of compatibility). From this point of view, Propositions 3.18 and 3.21 confirm that our definitions are correct. . This is the category of lax cones (with respect to the weight W ) for F with vertex E. Note that a lax cone W . This is given in turn by its components We define the category of σ-ω-cones Cones W σ,ω (E, F ) as the full subcategory of and every x ∈ W A. For a σ-ω-cone ξ with vertex E, W ξ + 3 B(E, F −) , we have a functor µ B = ξ * given by precomposition with ξ: is an isomorphism. We refer to the arrows ξ A (x), for A ∈ A and x ∈ W A, as the projections of the limit. As usual, an equivalent formulation of the universal property is that there is a representation µ B natural in the variable B (as in (3.6)), and ξ is recovered setting B = E, ξ = µ E (id E ). Remark 3.7. By reversing the direction of the 2-cells θ f in Definition 3.1 we have the notion of σ-ωop-natural transformation, which leads in turn to the definition of σ-ωop-limit. The fact that every σ-ω-limit in B is a σ-ωop-limit in B co , and vice versa, goes back to at least [ Then, as it is usual in the literature for the lax case, we can state and prove general results for σ-ω-limits, and use them for σ-ωop-limits too. A similar fact is true for σ-ω-colimits (that is, they are σ-ω-limits in the dual 2-category B op ), though we don't explicitly consider them in this paper. Remark 3.8. When W is the functor △1, constant at the unit category, we recover Gray's notion of conical σ-ω-limit. To show the natural isomorphism between the categories of cones for each notion, consider the well-known isomorphism 3. If Ω = Ω s , and Σ = A 0 , then {W, F } σ,ω is the strict limit {W, F } s . Ω-modifications and Ω-compatible limits We will now define the notion of Ω ′ -compatible σ-ω-limit, for Ω ′ another family of 2-cells of B. We begin with the conical case, for which it can be defined using the concept of Ω ′ -modification, and consider then the general weighted case. Also, given objects B, C of B, we denote by B Ω ′ (B, C) the category with objects B f −→ C and arrows f α =⇒ g such that α ∈ Ω ′ . or Ω ′ -compatible, if for every B ∈ B, the restriction of the isomorphism in (3.4 Remark 3.12. Consider the isomorphism π * in (3.4) and let π * ϕ = θ, π * ϕ ′ = θ ′ . Then, for each morphism of cones given by θ A The Ω ′ -compatibility of the limit means that, if all the components α A are in Ω ′ , then so is β (with the other implication always holding by the equation above). Remark 3.13. Consider as Ω ′ the families Ω γ , γ = s, p, ℓ of Example 2.4. Then every (conical) σ-ω-limit is Ω ′ -compatible. In fact, for γ = ℓ the condition is vacuous, and for γ = p, s this follows from the fact that (with the notation of the previous remark) invertible (resp. identity) α and β correspond via the isomorphism π * (note that a modification α is invertible, resp. the identity if and only if each of its components α A are so). / / F A of B, one for each A ∈ A and each x ∈ W A. We define Cones W σ,ω (E, F ) Ω ′ to be the category with the same objects as Cones W σ,ω (E, F ), that is σ-ω-cones, and arrows those morphisms ρ such that ρ A (x) is in Ω ′ for each A ∈ A and each x ∈ W A. We say that the limit E = {W, F } σ,ω is compatible with Ω ′ , or Ω ′ -compatible, if for every We leave to the reader the easy task of checking that, in the case W = △1, both definitions coincide (see also Proposition 3.21 below). We now examine carefully what it means to be Ω ′ -compatible for some limits with which we will work in Section 4. We refer to [7, §4] for details. We abuse language and refer to p as the projection of the inserter, though according to Definition 3.5 so are ξ b (0) and ξ b (1). The one-dimensional universal property is that for any (q, µ) as above there is a unique E h −→ L such that ph = q and λh = µ. The two-dimensional universal property is that for E h / / k / / L and ph β =⇒ pk such that (λk)(f β) = (gβ)(λh), there is a unique 2-cell h α =⇒ k with pα = β. Now, from Definition 3.14 and the computations above it is clear that the inserter is Ω ′ -compatible if, when β is in Ω ′ , so is α (note that if β = (ρ a ) * is in Ω ′ , then by the computations above so are (ρ b ) 0 and (ρ b ) 1 ). We refer to p as the projection of the equifier. Now, from the above it is clear that the equifier is Ω ′ -compatible if, when µ is in Ω ′ , so is λ. As noted in [1], this isomorphism implies immediately that any weighted lax limit can be expressed as a conical lax limit, but it also has the following interesting corollaries when restricting to "stricter" limits. In [11,Theorem 15] Street uses this to show that any strict weighted limit can be expressed a conical σ-ω-limit with Ω = Ω s (then in particular as a Gray's quasi-limit). In [4, §2.3] this procedure is slightly modified in order to prove that weighted σ-limits (given a family Σ of arrows of A) can be expressed as conical σ-limits. The expression of weighted limits as conical limits The notion of σ-ω-limit allows to easily show that these three results are instances of the same following proposition (considering respectively the items 1a, 3, 2 of Example 3.9), stating that any weighted σ-ω-limit can be expressed a conical σ-ω-limit with respect to the same family of 2-cells Ω. The indexing category for the conical limit is, as in the particular cases described above, the 2-category of elements El W of the weight. The diagram for the conical limit is given by the composition of the original 2-functor F with the projection El W ♦ W −→ A. The distinguished arrows of El W are those of the form (f, id), with f ∈ Σ. We denote this family of arrows by id Σ . We note that in [4] the family consisting of the arrows of the form (f, ϕ), with f ∈ Σ and ϕ an isomorphism is considered instead; below the proof we explain why this family works for the σ-case (that is with Ω = Ω p ) but not for this general case. Proof. For any other A G −→ Cat, we recall that the isomorphism (3.17) is given at the level of objects by the formulas, for η on the left side and θ on the right side, In particular, for G = B(E, F −) we consider the isomorphism between the categories of lax cones By the formula (η f ) x = θ (f,id) above, each (η f ) x is in Ω if and only if each θ (f,id) is, therefore if we consider the family id Σ of El W , the isomorphism above restricts to Cones W σ,ω (E, F ) ∼ = Cones σ,ω (E, F ♦ W ), thus finishing the proof. We note that, if f ∈ Σ and ϕ is an isomorphism, from the formula θ (f,ϕ) = η B (ϕ)(η f ) x it follows that θ (f,ϕ) ∈ Ω p (in other words that it is invertible) if (η f ) x is, but this may not be the case for an arbitrary Ω (which may not contain Ω p ). Remark 3.20. In the hypothesis of Proposition 3.18, the correspondence between the limit cones is given by the formulas present at the beginning of the proof of the proposition. If we denote these cones by ξ and π as in Definitions 3.3 and 3.5, we have in particular the formula ξ A (x) = π (x,A) for each A ∈ A, x ∈ W A, showing that the projections of each limit correspond. Proof. Recall that the isomorphism (3.17) is given at the level of arrows by the formula, for η α −→ η ′ on the left side and θ β −→ θ ′ on the right side, α A (x) = β (x,A) (for each A ∈ A, x ∈ W A). Then, by Definitions 3.11 and 3.14, the last isomorphism of the proof of Proposition 3.18 restricts to Example 3.22. We examine the result of Proposition 3.18 when applied to the inserter (see Example 3.15), as this will be relevant for our results (see Example 5.3). In this case the indexing category El W of the conical σ-ω-limit has three objects ( * , a), (0, b), (1, b) and three arrows from which all arrows can be obtained as identities and compositions, namely (u, id), (v, id), (id, ϕ) which are mapped respectively to f , g, id. This "conical inserter" is different to the following one that can be found in [6, I,7.10 2)]: there Gray considers the diagram {a In particular, we note that in the first case there are arrows of the family id Σ that are mapped to both f and g, but in the second one no arrow of Σ is mapped to f . Strict limits in the 2-categories of weak morphisms In this section we generalize the first limit lifting results of [3, §2] to the 2-categories of weak T -algebra morphisms, showing that these have products, inserters and equalizers when the base 2-category does (under some extra conditions, which are trivially satisfied in the cases for which these results were previously known). Their case is recovered setting Ω = Ω p . While these results are used in the rest of [3, §2] to deduce the lifting of lax and pseudolimits (using the construction of lax and pseudolimits from those three types of limits), we decided not to attempt a generalization of this deduction, since we will give in Section 5 a different, direct proof of a theorem regarding the lifting of σ-ω-limits, therefore in particular of lax and pseudolimits (see Remark 5.12). Though we don't use them for our main result of Section 5, the results of this section have their own relevance. First, setting Ω = Ω ℓ we have as a particular case of our Propositions 4.3 and 4.4 the results [9,Prop. 4.3,4.4], in the strengthened form of [9, §6]. We note that the proofs of these results, as they appear in [9], are substantially different from the ones of [3, Prop. 2.2,2.3], thus also to the ones here. Also, though our lifting theorem of Section 5 can also be applied to products, inserters and equifiers, when doing so we obtain in some cases a result which requires stronger hypotheses than the ones in this section, we examine this with detail for the inserter in Example 5.3. ω , if all the compositions p i z are ω-morphisms with respect to Ω ′ , then so is z. If Ω ′ = Ω s , we say "detect strictness". If Ω ′ = Ω p , we say "detect pseudoness". Proof. Our proof follows the same lines of [3,Prop. 2.1]. Since we also use the same notation we decided to omit some calculations that can be explicitly found in op.cit. If A = i∈I A i is an Ω-compatible product in K of a family of T -algebras, we have T A a −→ A the unique map with p i a = a i · T p. Then, as in [3,Prop. 2.1], (A, a) ∈ T -Alg Ω ω and p is strict. since the product is Ω-compatible (see Remark 3.12), given the 2-cells p i a.T h Note that, if the product is Ω ′ -compatible, h is in Ω ′ if all the q i are, giving the last assertion of the proposition. The rest of the proof (that is, the verification of the algebra axioms for (h, h) and the two-dimensional universal property) is exactly as in [3]. for which f is invertible and the inserter of (f, g) in K is Ω-compatible. The projection of the inserter is strict, and detects Ω ′ -ness for any family Ω ′ such that the inserter of (f, g) in K is also Ω ′ -compatible. If g is also invertible, we may replace in the proposition inserter by iso-inserter. Proof. The same remark that we made at the beginning of the proof of Proposition 4.2 applies here, with respect to [3,Prop. 2.2]. Given the inserter in K, (A, (p, λ)), we have the cone (q, µ) := (b · T p, (g · T p)(c · T λ)(f −1 · T p)) and therefore there exists a unique T A a −→ A such that pa = q and λa = µ. From here it follows, as in [3,Prop. 2.2], that (A, a) ∈ T -Alg Ω ω , that p is strict and that λ is an algebra 2-cell, so that (p, λ) is a cone in T -Alg Ω ω . Now, to prove the universal property, consider a cone (q, µ) in T -Alg Ω ω , with vertex D. Using the universal property in K, we have a unique D h −→ A such that ph = q, λh = µ. Now, a key observation for the proof (see [3,Prop. 2.2] for the computations) is that the equation (2.2) that expresses that µ is an algebra 2-cell is equivalent to the equation (λhd)(f q) = (gq)(λaT h), which expresses that q is a morphism between the cones (p · a · T h, λaT h) and (p · h · d, λhd), see Example 3.15. Then, since q is in Ω, and the inserter is Ω-compatible, there is a unique 2-cell a · T h h =⇒ hd in Ω such that ph = q. Note that, if the inserter is Ω ′ -compatible, h is in Ω ′ if q is, giving the last assertion of the proposition. The rest of the proof is exactly as in [3]. ) for which f is invertible and the equifier of (α, β) in K is Ω-compatible. The projection of the equifier is strict, and detects Ω ′ -ness for any family Ω ′ such that the equifier of (α, β) in K is also Ω ′ -compatible. Proof. The same remark of the proofs of Propositions 4.2, 4.3 applies here, with respect to [3,Prop. 2.3]. Given the equifier p in K, since f is invertible then b · T p equifies α and β and therefore we have a unique T A a −→ A with pa = b · T p. Then, as in [3,Prop. 2.3], (A, a) ∈ T -Alg Ω ω and p is strict. Now, to show that (p, id) is the equifier in T -Alg Ω , let (q, q) equify α and β with q ∈ Ω. By the one-dimensional universal property in K, there is a unique D h −→ A with ph = q. Now, since the equifier is Ω-compatible (see Example 3.16), given the 2-cell pa.T h q =⇒ phd there is a unique 2-cell a.T h h =⇒ hd in Ω with ph = q. Note that, if the equifier is Ω ′ -compatible, h is in Ω ′ if q is, giving the last assertion of the proposition. The rest of the proof is exactly as in [3]. Weak limits in the 2-categories of weak morphisms We now give what we consider our main result of this paper, a lifting theorem for 2-categories of ω-T -morphisms and conical σ-ωop-limits (that is, Gray's cartesian op-quasi-limits, see Remark 3.7). Various interesting limit lifting results follow as corollaries. Theorem 5.1. Let Σ be a family of arrows of A, and let Ω, Ω ′ Ω ′′ be families of 2-cells of K. We consider the family Ω consisting of those 2-cells α of T -Alg Ω ′ ω such that α ∈ Ω. We may say in this case, slightly abusing language, that the forgetful 2-functor U Ω ′ ω creates this type of σ-ωop-limits. The projections π A of this limit are strict, and jointly detect Ω ′′ -ness for any family Ω ′′ such that σωopLim We define a σ-ω-cone θ = (θ A , θ f ) with vertex T L, where θ A = aT (π A ) and θ f is given by the composition Note that for f ∈ Σ, π f ∈ Ω then by hypothesis so is T π f , and F f ∈ Ω also by hypothesis, therefore θ f ∈ Ω as desired. From the one-dimensional universal property of the limit The associativity and unit axioms for l, expressing that (L, l) is a T -algebra, follow from those of the F A using that the projections π A are jointly monic and the naturality of the unit and the multiplication of T . Then the π A are strict morphisms as desired, and the equality π f l = θ f is exactly the equation that expresses that the π f are algebra 2-cells, thus we have a σ-ω-cone in T -Alg Ω ′ ω . To show the one-dimensional universal property in T -Alg Ω ′ ω , consider another σ-ω-cone E By the universal property in K, there exists a unique h satisfying the first two of these equalities. Now, the equations that express the fact that each h f is an algebra 2-cell, namely are equivalent to the axiom expressing that h * e h A −→ θ * T h is a morphism of σ-ω-cones, that is a modification. Note that h * e = π * he, θ * T h = π * lT h, then by our Ω ′ -compatibility hypothesis we have a unique he h =⇒ lT h in Ω ′ satisfying π A h = h A as desired. Note that, if the limit is Ω ′′ -compatible, h is in Ω ′′ if each h A is, giving the last assertion of the theorem. The coherence conditions with respect to m, i for h follow from those of the h A using that, by the two-dimensional universal property of the limit in K, if a pair of 2-cells is equal after composing with all the π A , then they are equal. Now, for the 2-dimensional universal property in T -Alg Ω ′ ω , consider the following situation: Remark 5.4. The reason why [3, Prop. 2.2] doesn't follow from Theorem 5.1 using Proposition 3.18 is that we are trying to lift a σ-ω-limit which is taken with respect to Ω s to a 2-category of ω-T -morphisms with respect to Ω p , but Ω s is not included in Ω p , therefore our hypothesis that F (f ) is in Ω s for each f ∈ Σ won't necessarily hold. This obstruction vanishes in the case of Theorem 5.1 in which Ω ′ ⊆ Ω, and this allows the use of Proposition 3.18 to deduce the lifting of weighted σ-ω-limits: Proof. We can write a weighted σ-ωop-limit as a conical one using Proposition 3.18 (recall Remark 3.7). Since Ω ′ ⊆ Ω, all the arrows (g, g) of T -Alg Ω ′ ω satisfy g ∈ Ω, thus the hypothesis of Theorem 5.1 requiring F (f ) ∈ Ω is immediately satisfied. Note that Proposition 3.21 shows that the conditions of compatibility are preserved by the application of Proposition 3.18, and Remark 3.20 shows the correspondence between the projections, thus giving the last statement of the corollary. Remark 5.6. We could actually require in Corollary 5.5, instead of Ω ′ ⊆ Ω, a condition on the functor A F −→ T -Alg Ω ′ ω for {W, F } σ,ωop to be created by U Ω ′ ω , namely that F (f ) is in Ω for each A f −→ B in Σ such that W A = ∅. A more careful application of Proposition 3.18, expliciting the involved constructions, would allow to deduce this result from Theorem 5.1. Writing this with full detail would lead to results slightly stronger than our Corollaries 5.11, 5.13, for example we could deduce the lifting to any T -Alg γ of all limits which have a strict diagram (cf. [9,Prop. 4.1]). For the sake of simplicity, since we are unsure of how much value such a generalization adds, we have refrained here from doing so. Remark 5.7. We leave to the reader to write in its complete form the dual versions of Theorem 5.1 and Corollary 5.5, which state that the forgetful 2-functors T -Alg Ω ′ coω U Ω ′ coω − −− → K create the corresponding types of σ-ω-limits. These dual versions are equivalent to the corresponding results by Remarks 2.3 and 3.7 (or, can be seen to hold with dual proofs). We note that, if Ω ′ has only invertible 2-cells, by the last statement in Remark 2.3 we may consider the forgetful 2-functor T -Alg Ω ′ ω U Ω ′ ω −−→ K instead. Remark 5.8. In the situation of Theorem 5.1, if Ω is any of the three choices Ω s , Ω p , Ω ℓ , then the equality Ω = Ω makes sense and is easily seen to hold (noting for the Ω p case that if an invertible 2-cell satisfies (2.2) then so does its inverse). Also, for these choices of Ω, the hypothesis T (Ω) = Ω is immediately satisfied. If Ω ′ is any of those three choices, then the hypothesis of Ω ′ -compatibility is immediately satisfied (recall Remark 3.13 and Proposition 3.21). The same holds for Ω ′′ .	2017-04-04T08:00:41.000Z	2017-02-10T00:00:00.000	{ "year": 2017, "sha1": "5804c29dee401c56190930f8d6cf6d6826bb82f4", "oa_license": null, "oa_url": "http://arxiv.org/pdf/1702.03303", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "8034e681be8401112593f736411bf36cbf89a9c0", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics" ] }
226647260	pes2o/s2orc	v3-fos-license	Micro and small enterprises and human resources retention in a pandemic Micro and small businesses, especially information technology startups, as well as micro and small family businesses engaged in production using high technologies, and innovation, are a very important element of a country's economy. However, in conditions where the owner is usually the manager of the company and when there is no established human resources management sector or a designated position with responsibilities in the field of human resources management, the human management process itself is mostly informal, ad hoc, and subject to frequent changes. Human resource management problems are particularly noticeable in crisis situations such as the COVID-19 virus pandemic. This paper examines the attitudes (in the form of interviews) of managers and owners of small businesses in the field of information technology and high technologies towards the retention of employees in the COVID-19 virus pandemic in the first week of declaring a state of emergency in Serbia and in the first week of May, when the lifting of the state of emergency was announced. Although the sample is small, it can be concluded that small businesses were unprepared for this crisis, but that in terms of human resource management, especially when it comes to retention and rewarding employees, small businesses in the field of information technology have been able, because of the nature of their work, to overcome the crisis much easier than companies engaged in production (although they are utilizing high technologies). Introduction Micro and small enterprises are a very important element of the economy of every developed country, for example in the US more than half of the employees work for small companies (Dessler, 2013). Nonetheless, they are also becoming even more significant in the economy of developing countries (Druică, et al. 2017), especially when it comes to micro and small enterprises that apply modern technologies and innovations (Duplenko, 2013;Apanasovich, et al. 2017). Micro enterprises are defined by different authors as the ones with less than 10 or 20 employees and small enterprises are defined as the ones that employ between 10 (or 20) and 50 employees (Vukotić, Cvijanović, Aničić, 2014). In order for these companies to survive and thrive in the conditions of strong competition and globalization, one of the key elements are the people who bring added value with their knowledge, skills, and abilities in order to secure desired position in the market (Melo and Machado, 2013;Shafeek, 2016;Hitka, et al. 2018;Končar, et al. 2020). Various concepts of quality management state that human resources management is critical factor of any organization (Tomic, 2016), especially when it comes to small and medium enterprises (Cepel, et al. 2018), and some micro and small enterprises are starting to develop and implement human resources management policies and procedures (Melo and Machado, 2013). Micro and small enterprises are prevalent in the Republic of Serbia (Bobera and Bjekić, 2016; USAID. 2017), nevertheless there is not enough extensive research on the implementation of human recourses policies and practices in those companies in order to retain employees, or whether those policies and practices even exist, especially when it comes to the micro and small companies in the field of information technologies and high technologies. Natural disasters and epidemics can cause disruption of various business processes and can be harmful to people's health and lives. The problems for human resource management are particularly noticeable in the situations such as the COVID-19 virus pandemic. Therefore, risk management strategies have to include plans for safety and protection of personnel. It is essential for micro and small enterprises, where absence of only one key employee can cause a serious damage to business processes. The effect of pandemic on the economy on macro level has been researched in some extent in the shape of reports, reviews and analyses (Sanchez-Duque, et al. 2020), nevertheless the effect, especially long-term, of the pandemics on micro and small enterprises has yet to be researched more extensively. In this paper we shall examine the attitudes (in the form of interviews) of managers and owners of micro and small businesses in the field of information technology and high technologies. The participants have been interviewed soon after COVID-19 virus pandemic in March have started, namely the first week of declaring a state of emergency in the Republic of Serbia (Službeni glasnik, 2020) and in the first week of May, when the lifting of the state of emergency have been announced. The main focus has been given to retention and rewarding employees in pandemic. Human Resources management in micro and small enterprises Micro and small enterprises, especially in the field of information technologies and high technologies, depend heavily on the technology; nevertheless, new technological solutions can't function without the people (Jerônimo, 2013;Fejfarová and Urbancová, 2016), namely the individuals who have adequate knowledge, skills, abilities, other characteristics, and behaviours. Exceptional skills in human resources management in micro and small enterprises are the key for hiring and keeping the best talent. Almost typical, in these kinds of enterprises, the owner is often also a general manager, and CFO, but also a chief human resources manager. Therefore, it is necessary for some type of human resources management policies to be developed and implemented from the very beginning. A number of micro and small companies can opt for outsourcing this function (Dessler, 2013), or using HRM information systems (Wang, et al. 2016), however adopted policies and procedures have to be based on the company's mission and vision in order to be implementable. It is not unusual for micro, small, and even medium enterprises not to have human resources management department or a separate function. Usually companies larger than one hundred employees tend to employ one person in a human resources function (Dessler, 2013). A research from Check republic has shown that only about 25% of small and about 50% of medium enterprises (Fejfarová and Urbancová, 2016) include some form of human resources function. Nevertheless, micro and small companies in the field of information technologies or high technologies tend to treat their employees as human capital, and not resources (Apanasovich, et al. 2017) because of their importance for the success of a company. Nevertheless, one of the often neglected aspects of human resources management in micro and small enterprises is health and wellbeing of the employees (Thrive at Work Wellbeing Programme Collaboration, 2019). Micro and small enterprises in Serbia in most cases don't have a human resources function at all (Hanić, et al. 2016;Lazarević, et al. 2018), which is often essential for retaining the best employees. Significant decrease in official unemployment rate (10.4% for the 2019), according to the Statistical Office of the Republic of Serbia (2020), can be misleading, because it is very difficult to recruit and retain the professionals in the field of information technologies and high technologies who are the right fit for organization due to the brain drain and considerable opportunities for informal online employment, especially for the professionals in the field of information technologies (Culkin, Simmons 2018). Sometimes, start-ups are not able to provide the salary, bonuses, and overall working environment that can be attractive for the best candidates, the ones with the experience and interns alike. Human resources management in pandemic Even though pandemic incidences can't be anticipated, after SARS and H1N1 epidemics, same as epidemics of Ebola and Zika viruses, it is necessary to include policies and procedures for the possible epidemics and pandemics response into risk management and human resources management strategies (Caligiuri, et al. 2020). It is of the essence for those policies and guidelines to include the human resources management policies in order to retain talented individuals, because even though small scale epidemics may not be harmful for large enterprises' business processes, the absence of employees in micro and small enterprises, especially if the absent employees are holding key job positions that cover various responsibilities, can cause a serious damage to a company (Tan and Takakuwa, 2011). Coronavirus Disease 2019 (COVID-19) caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) has been first diagnosed in December 2019 in the city of Wuhan in Hubei province in China and has become globally known from the beginning of January 2020. As the virus spread from China, across Japan, South Korea, and Singapore to US, Western, and Eastern Europe, the outbreak has affected more than one hundred countries along with the Republic of Serbia from the beginning of March 2020. World Health Organization (WHO) declared pandemic on March 11 (Lenzen, et al. 2020). Besides the health issues, one of the main problems for the micro and small enterprises is loss of jobs in service and retail sector (da Silva Costa, 2020), because of widespread lockdowns in order to contain the epidemic (Musinguzi and Oppong, 2020) with serious consequences for the retention of employees for the companies involved in sophisticated production. Also, for these companies, the question of post-Covid economic survival is becoming of the essence (Baghiu, 2020; Caligiuri, et al. 2020). Many countries have passed government led programs of aiding the most affected industries, especially through wage subsidies (Joo-Cheong, 2020). Similarly in the Republic of Serbia, Government (Vlada Republike Srbije, 2020) has issued a set of economic measures intended to help private sector coping with the financial consequences of the COVID-19 pandemic. The package has included salary tax and social security contributions on salaries, salary compensations, and personal salaries of entrepreneurs (Vlada Republike Srbije, 2020), especially important for micro and small enterprises. Research The scope of study and data collection Objectives The aim of the empirical part of the paper is to analyze the attitudes of owners and top managers of micro and small enterprises in the field of information technologies and high technologies (sophisticated automated production) toward their ability to retain and reward the employees during (and immediately after) the Covid-19 pandemic. The interviews have been held from 18 th to 20 th March (the beginning of the state of emergency) and from 7 th to 8 th May, 2020 when abolition of the state of emergency has been announced (Narodna skupština, 2020), through semi-structured phone interviews and the answers have been transcribed by the authors. Survey design The interviewees have been given two groups of statements with two possible answers (Yes/No) questions. The first group of statements has been developed to investigate whether a company has: 1. The business plan for the 2020 that includes human resources plan; 2. Human resources management function; 3. Human resources management position; 4. Human resources management policies, procedures, and plans; 5. Risk management policies, procedures, and plans; 6. A plan for emergency situations regarding the employees included in the risk management policies, procedures, and plans. The second group of statements has been developed to investigate whether a company has: 1. The ability to continue the business plan for the 2020; 2. The ability to retain all the employees; 3. The ability to retain the bonuses for the employees 4. The ability to retain regular wages for the employees 5. The ability to retain essential (core) employees during/after pandemic 6. The ability for online business (core bussiness) 7. The ability for enabling the employees to work online during/after the pandemic (administration and management) Also, the participants have been able to state if their company would be willing to accept some kind of financial assistance from the governement in order to retain employees (March) or if it has accepted financial government's aid package (May). Furthermore, the participants have been given the possility to explain their willinness to accept governments's financial aid. The participants have also been given the ability to present additional remarks on the issues they have considered important, if willing. Description of the sample For the purpose of this study, nineteen owners/general managers/chief executive officers of various micro or small companies (fewer than 20 employees) in the field of information technologies and production (using high technologies) have been interviewed. The interviewees have been the owners of the companies that have been active from one to twenty years. Ten of the participants have been from the field of information technologies and nine from the field of high technologies. The demographical data of the sample are given at Table 1. Results The first set of statements has been given only in the initial interviewing in March. In regard to business plans all of the interviewees have stated that their companies have a business plan developed for the year 2020. In IT industry most of the participants (80%) have stated that they have a human resources management function, but none has stated that they have a human resources management position. It is similar for high technologies, where the most of the interviewees (66.7%) have stated that they have a human resources management function, but, again, none has stated that they have a human resources management position. Most of the companies in the field of information technologies (70%) have written human resources policies, procedures and plans; nevertheless, the most of high technology companies (88.9) do not have written human resources plans, policies and procedures. The most of the companies in both fields don't have risk management plans, policies, and procedures or plans for emergency situations in regard to the employees (80% and 66.7% respectively), most probably because those plans have not been legally bound until recently. Nevertheless, it can be seen that more high technologies companies than information technologies companies include those plans, because of the nature of their business. The results can be seen at Table 2. In regard to the ability to retain and reward the employees, all of the companies in the field of information technologies in both periods have stated that they have the ability to continue their business plans for 2020. Regarding the high technologies companies, the situation has been entirely different. Only 22.2% of the companies in March and only one company (11.1%) in May have believed in their ability to continue business plans for 2020. In March, all companies in information technologies field have believed to be able to retain all the employees, with only one that changed the statement in May. In high technologies, 44.4% in March already have believed not to be able to retain all the employees and that percentage decreased even more in May (22.2%). The most of the information technologies companies (80%) have believed in their ability to retain the bonuses with slight May decrease (70%). And the most of those companies have believed to be able to retain regular wages, namely 90% in May and 90% in May. Completely different situation has been for high technologies field. Only one interviewee in March (11.1%) have believed that they would be able to retain the bonuses, and in May, all the participants from the field of high technologies have believed that they won't be able to retain bonuses. It is similar for regular wages, 22.2% of interviewees in March and only one in May have believed to be able to retain regular wages. Understandably, the complete opposite attitudes have been expressed for the online (core) business possibilities, because none of the high technologies companies is able to have exclusively online core business, but all information technologies' companies are. Similarly, in regard to possibility of management and administration to be able to function online there is a discrepancy. All of the information technologies' participants have believed that their companies are able to continue managerial and administrative functions completely online (both in March and May), it has been different for high technologies companies, where in March 44.4% of participants have stated that they would be able to continue managerial and administrative function online in March, and only one in May. Some of the participants have stated that that their administrative and managerial function requires a lot of physical presence. The results can be seen at Table 3. The attitudes toward the acceptance of possible financial governments' aid in March in May have not been changed. None of the participants have stated that they would apply (or has applied) or government's aid. The results can be seen at Table 4. Because none of the participants have stated that they would be willing to accept the possible government aid, the reasons for that thinking have been further investigated in May interviews. There is a huge gap in the reasons for declining the government's aid package. Information industry representatives have stated that regarding the human resources they are quite capable to overcome the crisis. However, the representatives of high technologies have stated that the government's aid won't help them, because they wouldn't be able to continue even in those circumstances. The results can be seen at Table 5. Limitations and scope of further research This research includes the sample that is small, further research is needed and this very important issue calls for larger sample, survey and operation of various statistical methods. Conclusion Most of the micro and small companies don't have organized human resources management sector or a designated position with responsibilities in the field of human resources management. Moreover, those companies also don't have risk management or human resources management plans that include retention of the employees in crisis situations such as a pandemic. According to the owners and general managers of micro, and small businesses in the field of information technologies, there is different situation in coping with the retention of the employees between the two industries; even both are based on high end information technologies, high technologies, and innovation. Micro and small companies in the field of information technologies have been, because of the nature of their work, less affected by the disruptions caused by pandemic and the state of emergency, and the micro and small companies in the field of high technologies have not been able to cope with the effects of pandemic because of disruptions in supply chain and loss of a market. Although the sample is small, it can be concluded that micro and small businesses have been unprepared for this crisis, but that in terms of human resource management, especially when it comes to retention and rewarding employees, micro and small businesses in the field of information technologies have are able to overcome crises much easier than companies engaged in production (although they are utilizing high technologies) and to retain and reward the best talent. Further research is needed with much larger sample, surveys and statistical analysis to further analyse this important topic.	2020-10-29T09:07:13.364Z	2020-01-01T00:00:00.000	{ "year": 2020, "sha1": "7e8f7cd8f06cb36b9a1f809a5f67e7caafeacf54", "oa_license": null, "oa_url": "https://scindeks-clanci.ceon.rs/data/pdf/2466-4693/2020/2466-46932002077I.pdf", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "0e2c547a5179ff50a8c76fe33475c2109b54cbfe", "s2fieldsofstudy": [ "Business" ], "extfieldsofstudy": [ "Business" ] }
251282339	pes2o/s2orc	v3-fos-license	Robust optimization of VMAT for prostate cancer accounting for geometric uncertainty Abstract The aim of this study was to propose optimal robust planning by comparing the robustness with setup error with the robustness of a conventional planning target volume (PTV)‐based plan and to compare the robust plan to the PTV‐based plan for the target and organ at risk (OAR). Data from 13 patients with intermediate‐to‐high‐risk localized prostate cancer who did not have T3b disease were analyzed. The dose distribution under multiple setup error scenarios was assessed using a conventional PTV‐based plan. The clinical target volume (CTV) and OAR dose in moving coordinates were used for the dose constraint with the robust plan. The hybrid robust plan added the dose constraint of the PTV‐rectum to the static coordinate system. When the isocenter was shifted by 10 mm in the superior–inferior direction and 8 mm in the right‐left and anterior directions, the doses to the CTV, bladder, and rectum of the PTV‐based plan, robust plan, and hybrid robust plan were compared. For the CTV D99% in the PTV‐based plan and hybrid robust plan, over 95% of the prescribed dose was secured in all directions, except in the inferior direction. There was no significant difference between the PTV‐based plan and the hybrid robust plan for rectum V70Gy, V60Gy, and V40Gy. This study proposed an optimization method for patients with prostate cancer. When the setup error occurred within the PTV margin, the dose robustness of the CTV for the hybrid robust plan was higher than that of the PTV‐based plan, while maintaining the equivalent OAR dose. INTRODUCTION The volume required for target tumors is determined by the radiation therapy plan. 1, 2 The clinical and anatomical concepts of tumors define the gross tumor volume and clinical target volume (CTV). To administer a sufficient dose to the CTV, a margin must be added. Internal margin (IM) is a range that includes uncertainty due geometric margin in the moving coordinates, determines whether a CTV receives a prescribed dose. 5,6 In heavy ion and proton therapy, the main idea is to have a robust plan. 5,7,8 Because the dose distribution changes significantly due to the use of the Bragg peak and slight changes in density in proton radiotherapy, it is vulnerable to position fluctuations. Because the dose distribution in heavy ion and proton radiotherapy can fluctuate significantly, the idea of PTV is not effective. 9 Thus, rather than the PTV margin, dose robustness should be discussed in terms of dose distribution. This robust planning technique has not been widely used in photon radiotherapy. At the research level, there are some reports on the robust planning of photon radiotherapy. [10][11][12][13] When compared to conventional radiotherapy, a robust plan for volumetric modulated arc therapy (VMAT) in the lung secured the tumor dose while reducing the dose to normal tissues. 9 Mahmoudzadeh et al. showed that a robust plan with intensity-modulated radiotherapy (IMRT) reduced the dose to the heart when compared to the dose for breast cancer using conventional IMRT. 11 For glioblastoma, the target dose was equivalent to that of conventional IMRT, and the cerebral cortex was better preserved than in conventional IMRT. 14 These studies evaluated the robust optimization for a target that was not near the organ at risk (OAR). For prostate cancer, even if a setup error occurs, the target dose is sufficient, and the bladder and rectum suppress the dose as much as possible. In the clinical setting, daily cone beam computed tomography analysis has revealed a decrease in the dose of CTV, which has been linked to an increase in the number of prostate cancer recurrences. 15 Jin et al. reported that the assessment of the dose uncertainty in prostate IMRT leads to tumor control and risk reduction, which requires treatment planning that takes dose uncertainty into account. 16 Recently, a novel function was developed in RayStation (RaySearch Medical Laboratories AB, Stockholm, Sweden). It is capable of incorporating uncertainty into the planning optimization process and generating a robust plan. 17 The robust plan has the advantage of identifying the best scenario for the trade-off between dose robustness and OAR constraints. 18 However, the dose constraint for the robust plan is yet to be determined. Conventionally, the beam arrangement and optimization were performed in such a way that the prescription dose in the static coordinates (PTV-based plan) was delivered to the PTV. 19 Therefore, between PTV-based and robust plans, the concept of optimization differs. The optimal dose constraint should be determined for a robust plan. The current study focused on the prostate adjacent to the OARs. This study aimed to determine the treatment plan method with a robust plan by evaluating the robustness of a moving target with a PTV-based plan. Moreover, the robust and conventional PTV-based plans for the target and OARs were compared. PTV-based plan The planning CT scans were conducted with a slice thickness of 2.5 mm on LightSpeed16 (GE Healthcare, Chicago, IL, USA). RayStation Ver.6 was used for contouring, treatment planning, and dose evaluation. A region of interest of the whole prostate was generated using magnetic resonance imaging. Radiation oncologists used our Institution's protocol to create the contouring CTV, which was used in clinical patients. CTV was defined as the prostate and half seminal vesicles. The PTV margin was added to the CTV -10 mm to the superior and inferior, 8 mm to the anterior, right, and left, and 6 mm to the posterior. The target for the prescription was the PTV minus the part that overlapped with the rectum. The rectum and bladder were described as the OARs. The region of the rectum was defined as a 4-mm rectal wall. The region of the rectum overlap was defined as the area where the entire rectum and PTV overlapped. The bladder was defined as the 4-mm bladder wall. A total of 78 Gy in 39 fractions were prescribed for the PTV-rectum. The dose constraints used in the optimization parameters are listed in the appendix. A TrueBeam linear accelerator (Varian Medical Systems, CA, USA) was used in this study. A photon beam energy of 10 MV was chosen for this study. The VMAT plan was created with a partial arc of 210-150 • . The collimator angle was set to 10 • . The center of the CTV was defined as the isocenter. Table S1 summarizes the optimization parameters. Robust plan A robust plan was generated from the dose distribution of multiple setup error scenarios using RayStation Ver.6. The robust optimization aims at minimizing the expected value of the worst-case distribution. 17 The number of scenarios scales up in proportion to the degree of uncertainty and the number of shift directions. These scenarios include a no setup error scenario, endpoint setup error scenario, and intermediate setup error scenario in the right-left, inferior-superior, and posterioranterior directions. The setup error, which was assumed to occur within the PTV margin, was determined to be the scenario. 20 Namely, the CTV dose with the setup error in the PTV-based plan, which was used for the prescribed dose for the robust plan, was evaluated. The dose distribution in the moving coordinates of the PTVbased plan determined the dose constraint in the robust plan. For the robust plan in the moving coordinates system, the robust optimization function was used for CTV. Figure 1 shows illustrations of CTV in the PTV-based plan and CTV in the robust optimization planning. The robust plan optimization parameters are listed in the appendix. The robust optimization function of the robust plan was used only for the CTV. The optimization parameters of the CTV and OAR in the robust plan were determined by the dose of CTV and OAR when the PTV-based plan was shifted by 8 mm in the right, left, and anterior directions and 10 mm in the inferior and superior directions. The median values of D 1% and D 99% of the CTV in all directions were used for target optimization in the robust plan. The median values in all patients with average V 75Gy , V 70Gy , V 60Gy , and V 40Gy of the rectum, and V 75Gy , V 70Gy , V 60Gy , and V 40Gy of the bladder were used as optimization parameters in the robust plan for the OAR. The optimization parameters are summarized in Table S2. The dose constraints of robust optimization were established to be equivalent to the rectum and bladder dose of the PTV-based plan. We recalculated the PTV-based plan with multiple setup error scenarios and determined the rectal and bladder dose constraints from the calculated dosevolume histogram. In this study, we did not intentionally normalize the robust plan with PTV. The common normalization was not suitable because the PTV-based plan was delivered directly to PTV, and the robust plan using robust optimization algorithms was based on CTV. 9 Hybrid robust plan The hybrid robust plan was generated using RayStation. For the target optimization parameters, the robust plan used the CTV in the moving coordinates, and the PTV-based plan used the PTV in the static coordinates. The hybrid robust plan was defined as the one that utilized both the PTV in the static coordinates and the CTV in the moving coordinates. As shown in Table S3, the optimization parameters of the PTV-rectum used in the PTV-based plan were added to the optimization parameters and setup error scenarios of the robust plan. Plan evaluation For the PTV-based plan, robust plan, and hybrid robust plan, the isocenter was shifted by 10 mm in the superiorinferior direction and 8 mm in the right-left and anterior directions. The range of the shifted isocenter was within the PTV margin. In the moving coordinate system, the isocenter is moved to three axes in the three axes defined as lateral (right-left), longitudinal (superiorinferior), and vertical (anterior-posterior). The target and OAR dose were evaluated with the dose distribution calculated with the moving coordinate. For each patient, the doses of the CTV, rectal wall, and bladder wall were calculated. The rectal and bladder walls were both 4-mm thick. The points of evaluation were D 99% and D 1% of the CTV, V 75Gy , and V 70Gy , V 60Gy , V 40Gy of the rectum, V 75Gy and V 70Gy , V 60Gy , V 40Gy , of the bladder. 21-23 Statistics A t-test was used to compare the difference between the hybrid robust plan and the PTV-based plan or the robust plan in terms of dose robustness. The statistical significance threshold was set at p < 0.01 to be considered statically significant. Figure 2 shows a comparison of the D 99% of the CTV between the PTV-based robust plan and the hybrid robust plan when the isocenter shifted within the PTV margin. For the robust and the hybrid robust plans, the prescription was normalized at the mean dose of the CTV (CTV Dmean) for each plan, with setup error in the PTV-based plan. The CTV Dmean ± 2SD of the robust plan, the hybrid robust plan, and the PTV-based plan were 7930 ± 131 cGy, 7962 ± 75 cGy, and 7866 ± 69 cGy. The hybrid robust and the robust plans had a significantly higher CTV Dmean than the PTV-based plan. There was no significant difference in CTV Dmean F I G U R E 2 Comparison of clinical target volume (CTV) D 99% between the hybrid robust plan, the robust plan, and the planning target volume (PTV)-based plan between the hybrid robust plan and the robust plan. In the CTV D 99% , the hybrid robust plan was significantly higher than the robust plan in all directions and significantly higher than the PTV-based plan except in superior and posterior directions. The robust plan was not significantly higher than the PTV-based plan in CTV D 99% . On the posterior side, the PTV margin was lower to reduce the rectal dose, and we only provided reference information. Except for the inferior direction, over 95% of the prescribed dose was secured in all directions in the PTV-based plan. The mean dose of CTV D 99% in the inferior direction was 7097 cGy, which was less than the prescribed dose of approximately 9%. In the hybrid robust plan, over 95% of the prescribed dose was secured in all shift directions. When the isocenter shifted within the PTV margin and the CTV D 99% was compared between the hybrid robust plan and the robust plan,the robust plan had less than 95% of the prescribed dose in all directions. The robust plan removed the PTVrectum from dose constraints. Obviously, in the hybrid robust plan that included the PTV-rectum in dose constraints, the CTV D 99% was higher. The CTV dose tends to be significantly reduced when the robust plan has a setup error. Figure 3 shows a comparison of the D 1% of the CTV between the PTV-based, robust, and hybrid robust plans when the isocenter shifted direction. Across the board, D 1% of CTV in the hybrid robust plan was significantly higher than in the PTV-based plan. The robust plan was significantly higher than the PTV-based plan except in the superior direction. There was no significant difference between the hybrid robust plan and the robust plan. When the hybrid robust plan shifted 10 mm in the inferior direction, the D 1% of CTV was the highest at 8378 cGy. The comparison of the mean values and standard errors of the rectum V 75Gy , V 70Gy , V 60Gy , and V 40Gy for the PTV-based plan, robust plan, and hybrid robust plan F I G U R E 3 Comparison of clinical target volume (CTV) D1% between the hybrid robust plan, the robust plan, and the planning target volume (PTV)-based plan is shown in Figure 4. The plotted dash line for each metric indicates the clinically acceptable level at no setup error scenarios. [21][22][23] The rectal dose increased when there were setup errors in the anterior and superior directions. The hybrid robust plan has a significantly higher dose than the PTV-based plan for the V 75Gy of the rectum with no setup error and 8 mm to the right. There was no significant difference between the PTVbased plan and the hybrid robust plan for rectum V 70Gy , V 60Gy , and V 40Gy . The rectal dose in all metrics of the robust plan did not have a significantly higher dose than the PTV-based plan. The comparison of the mean values and standard errors of the bladder V 75Gy , V 70Gy , V 60Gy , and V 40Gy between the PTV-based plan, robust plan, and hybrid robust plan is shown in Figure 5. The bladder dose increased when there was a setup error in the inferior direction. There was no significant difference between the PTV-based plan and the hybrid robust plan for bladder V 75Gy , V 70Gy , V 60Gy , and V 40Gy . The robust plan was significantly lower than those in the hybrid robust plan and the PTV-based plan in V 75Gy , V 70Gy , V 60Gy , and V 40Gy . DISCUSSION For conventional treatment planning assessment, PTV coverage in static coordinates was regarded as important. However, the concept of PTV has limitations. 5 The target received the ideal dose distribution without taking into account the OAR dose because the PTV was assumed to be not equally conformal on all sides of the CTV. In other words, the dose distribution is assumed to be completely homogeneous within the PTV and to have a constant spread in all directions outside of it. The prostate is not spherical in clinical radiotherapy for prostate cancer, and it is adjacent to risk organs like the bladder and rectum. To reduce the exposure to OAR as much as possible, the dose distribution with VMAT does not have a uniform spread outside the PTV. In this study, we used clinical patients and used the shifting isocenter to evaluate the dose of the target and OAR in the moving coordinates. At the prescribed dose, the PTV-based plan showed a dose reduction of approximately 9%. The difference in the geometrical shape concept and the real shape of the prostate of a patient being administered a dose for prostate cancer affects the extent of dose reduction. The CTV dose was significantly reduced in the inferior direction even when the isocenter position was shifted within the PTV margin. For the hybrid robust plan, a dose of 95% or more of the prescribed dose was delivered to the CTV. This indicates that the hybrid robust plan is maintained at a higher dose than the PTV-based plan. The hybrid robust plan has higher maximum and rectal doses than the PTV-based plan. However, for RTOG 0126 and other reports, the rectal dose with the hybrid robust plan met the dose constraint of the dose standard. 23,24 Robust optimization can optimize one target with setup error within one scenario. Namely, the uneven expansion of the CTV in the direction of the respective organs of the rectum and bladder cannot be considered. Future research will determine if robust optimization improves multiple scenarios. The clinical outcomes of the PTV-based plan were promising, and robust hybrid plans were designed to keep the PTV-based plan. The goal of this research was to improve dose robustness by reducing the worst-case scenarios such as a significantly lower target dose or a significantly higher dose of the OAR for the PTV-based plan. The PTV-based and robust plans have different prescription targets and optimization methods. Therefore, they were not normalized on purpose. Liang et al. study reported that both the PTV-based and robust optimization cannot be the same, and the traditional clinical prescription for the robust plan is inappropriate. 9 Therefore, in this study, for the dose constraints of CTV and OAR in the robust plan, we used conventional clinical data. When setup errors occurred, the doses of CTV and OAR were calculated. Moreover, the dose constraint of the PTV-rectum supports the robustness of the CTV dose. In the hybrid robust plan, the PTV-rectum was F I G U R E 5 Comparison of bladder dose between the hybrid robust plan, the robust plan, and the planning target volume (PTV)-based plan. (a) V 75Gy , (b)V 70Gy , (c)V 65Gy , and (d)V 40Gy required to ensure coverage of the CTV dose, despite variations in the setup position. The hybrid robust plan helped to avoid dose reduction in the robust plan. The robust optimization in RayStation is based on minimax optimization, which considers the optimization functions that are robust in a worst-case scenario. 17 With a similar number of scenarios as in the current study, Byrne et al. demonstrated that a robust plan is resistant to setup errors. 10 The proposed hybrid robust method improves the target coverage while maintaining the nonsignificant difference in OAR dose in most of the random setup errors as compared with the conventional robust plan. From above, the proposed dose constraints and scaling of the prescription dose can be proposed as a practical method for determining the dose constraint of robust optimization. CONCLUSIONS This study proposes a robust optimization method for prostate VMAT. The dose robustness of the CTV when the setup error occurred within the PTV margin was higher for the hybrid robust plan than for the PTVbased plan while maintaining equivalent OAR dose. The current study suggests that a robust optimization can reduce dose uncertainty due to setup errors. C O N F L I C T O F I N T E R E S T The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported. AU T H O R C O N T R I B U T I O N S TW and DK conceived and designed the study, and write the manuscript. TW performed data collection, data analysis and interpretation of results. WT, DK, YM, TN and YN helped interpret the data and write the manuscript. DATA AVA I L A B I L I T Y S TAT E M E N T The data that support the findings of this study are available from the corresponding author upon reasonable request.	2022-08-04T06:17:07.437Z	2022-08-03T00:00:00.000	{ "year": 2022, "sha1": "f62cdecf19b0a553eb67e447571843c98aff2aba", "oa_license": "CCBY", "oa_url": null, "oa_status": "CLOSED", "pdf_src": "PubMedCentral", "pdf_hash": "fe05977490b0f8a67713a8c44aea0cce89aa89fc", "s2fieldsofstudy": [ "Medicine", "Engineering" ], "extfieldsofstudy": [ "Medicine" ] }
7576751	pes2o/s2orc	v3-fos-license	Association between maternal education and objectively measured physical activity and sedentary time in adolescents Background Investigating socioeconomic variation in physical activity (PA) and sedentary time is important as it may represent a pathway by which socioeconomic position (SEP) leads to ill health. Findings on the association between children's SEP and objectively assessed PA and/or sedentary time are mixed, and few studies have included international samples. Objective Examine the associations between maternal education and adolescent's objectively assessed PA and sedentary time. Methods This is an observational study of 12 770 adolescents (10–18 years) pooled from 10 studies from Europe, Australia, Brazil and the USA. Original PA data were collected between 1997 and 2009. The associations between maternal education and accelerometer variables were examined using robust multivariable regression, adjusted for a priori confounders (ie, body mass index, monitor wear time, season, age and sex) and regression coefficients combined across studies using random effects meta-analyses. Analyses were conducted in March 2014. Results Adolescents of university educated mothers spent more time sedentary (9.5 min/day, p=0.005) and less time in light activity (10 min/day, p<0.001) compared with adolescents of high school educated mothers. Pooled analysis across two studies from Brazil and Portugal (analysed separately because of the different coding of maternal education) showed that children of higher educated mothers (tertiary vs primary/secondary) spent less time in moderate to vigorous PA (MVPA) (6.6 min/day, p=0.001) and in light PA (39.2 min/day: p<0.001), and more time sedentary (45.9 min/day, p<0.001). Conclusions Across a number of international samples, adolescents of mothers with lower education may not be at a disadvantage in terms of overall objectively measured PA. INTRODUCTION Socioeconomic position (SEP) is an important correlate of health and well-being. Studies show that in adults from developed countries, low SEP, measured through household income and education, is related to poor health and greater all-cause mortality. 1 2 Likewise, populations from developing countries who are socioeconomically disadvantaged in terms of income or education, tend to fare worse in terms of non-communicable diseases, 3 although some epidemiological evidence suggests that the early adoption of unhealthy behaviours by newly socioeconomically advantaged populations may lead to an inverse association between education and health in some countries. 4 Children growing up in low SEP families in developing countries are more likely to be overweight 5 6 and have greater risk for cardiovascular disease 2 7 and poor psychological health 8 than children of a higher SEP, although the association may differ by country of origin. 9 Socioeconomic variation in physical activity (PA) is important as it may represent an intermediary pathway by which SEP leads to poor health. In PA research, however, SEP is rarely used as a primary variable of interest, and is more frequently included to account for potential confounding effects. Within the relatively small body of literature that has directly examined associations between PA and SEP, findings are equivocal and tend to focus on middle/high income and/or developed countries. 10 11 A systematic review (which included papers from the USA, Asia, Canada, Oceania, Brazil, Europe (and 1 study that included 32 countries)) suggested that higher SEP was associated with higher levels of PA in adolescents. 11 However, 42% of studies reported no association or a negative association. Reasons for these equivocal results are that previous studies 10- 13 have used (1) varying indicators of SEP, (2) subjective, self-reported measures of PA, which are known to have inherent biases, 14 (3) varying domains (eg, active travel, leisure time) of PA. That said, a more consistent association has been shown between SEP and indicators of sedentary behaviour, such as selfreported television watching in Europe and the USA. 5 10 15-18 Fewer studies have examined the association between objectively measured total sedentary time and SEP. Objective measurement tools such as accelerometers, more accurately quantify PA and total sedentary time than self-reporting methods. 14 However, even among studies using objective measures, findings on the association between SEP and PA 17 19-22 and/or sedentary time 15 24 pooled and analysed raw accelerometer data using standardised methods to create comparable outcome variables across studies from Europe, Australia, Brazil and the USA. In addition, separate indicators of maternal education were pooled and recoded to create standardised maternal education categories across studies. Therefore, the ICAD affords the opportunity to investigate, for the first time, cross-sectional associations between maternal education, PA and time spent sedentary across a large number of international samples using standardised SEP and accelerometer data in a meta-analytical approach. Study design The ICAD (http://www.mrc-epid.cam.ac.uk) has been described in detail elsewhere along with dates of data collection. 24 In brief, the eligibility criteria for inclusion in the ICAD were (1) PA data in the form of raw accelerometer (.dat) files from a version of a waist-worn Actigraph accelerometer (eg, 7164, 71256, GT1M1) on children aged 3-18 years, (2) accompanying data, at a minimum, of gender, age and measured height and weight. Formal data sharing agreements were established, and all partners consulted with their individual research boards to confirm sufficient ethical approval had been attained for contributing data. Participants For the present study, the analytical sample (n=12 770) was restricted to children aged 10-18 years. This age range was chosen as it yields the largest number of participants with maternal education information. Data were taken from 10 studies, across eight countries: Australia ( Maternal education Owing to the variation in national education systems and categories used to record educational level (see online supplementary table S1), raw maternal education variables were standardised to enable comparisons across studies. To maintain meaningful interpretation, the studies were divided into two groups for analysis, and maternal education was standardised separately for each group. For eight of the studies, the raw education variables were recoded into the following three categories: (1) 'high school'-no education, or completed some/all of high school, (2) 'college'-completed some/all of college education (including vocational training), (3) 'university'-completed some/all of a university degree (including graduate school). The reason for selecting these categories (rather than including a 'not completed high school' category, commonly seen in other literature) is because three studies did not collect data on not completing high school, and when they did few were placed into this category. Where data was collected on not completing high school (5 studies), different criteria/cut-offs were used. The raw education variable for the remaining two studies (Pelotas study of Brazilian children and The Portugal European Youth Heart Study (EYHS) of children from Madeira) recorded school attainment in years of education, and thus, no distinction could be made between vocational and university training. These studies were analysed separately on an educational scale defined as (1) primary school education and (2) secondary school/tertiary education. Online supplementary table S1 also indicates the number of years of education for each education category by study. Assessment of PA and sedentary time Detailed description of the assessment of PA is available elsewhere. 24 Briefly, all available accelerometer data from the ICAD project were reanalysed, to provide comparable PA and sedentary outcomes across studies, using specifically developed and commercially available software (KineSoft, V.3.3.20, Loughborough, UK; http://www.kinesoft.org). Data files were reintegrated to 60 s epochs and non-wear time was defined as 60 min of consecutive zeros allowing 2 min of non-zero interruptions. 25 All children with ≥ 1 day with ≥500 min of wear time between 7:00 and midnight were included, in keeping with previous analyses. 26 A more thorough discussion of the appropriateness of using a single day of wear is available elsewhere. 27 Raw accelerometer-derived PA data were expressed as average counts per minute (CPM; ie, total counts per valid day, divided by monitor wear time per day). Time spent sedentary was defined as all minutes <100 CPM, 28 and MVPA time as minutes >3000 CPM, 28 29 corresponding to about 4.6 metabolic equivalents. 28 Anthropometry Height and weight was measured using standardised techniques within studies. Body mass index (BMI) was calculated as weight (kg)/height (m) 2 and participants were categorised into normal weight, overweight and obese groups according to age-specific and sex-specific thresholds. 30 Statistical analysis Descriptive results are expressed as mean and SD for continuous variables and percentages for categorical variables. Two-way analysis of variance (ANOVA) was used to explore possible interactions between sex and maternal education in relation to each PA outcome. Associations between CPM, light activity, moderate-to-vigorous PA (MVPA), and sedentary time were analysed using Pearson correlation. Robust linear regression was used to estimate the mean difference in PA outcomes between the maternal education categories within each study. Models were adjusted a priori for age, sex, season and monitor wear time, with the exception of CPM models, in which wear time is inherent in the derived variable. A binary classification was used as an indicator of season (based on daylight). Data collected in the Northern hemisphere were classified as either summer (data collected May-October) or winter (data collected November-April), and inverted for Southern hemisphere data. Reprocessing the raw accelerometer data allowed measurement of the dependent and independent variables on the same scales, and thus, comparison of mean differences in PA between maternal education categories. Mean differences of the studies were combined to estimate an overall mean difference (ie, a combined estimate), and 95% CI between maternal education categories for each accelerometer outcome in turn by using random effects meta-analysis models weighted by the inverse of the variances ( precision) of the mean differences. Heterogeneity among the study estimates was assessed by visual inspection of forest plots and by the I 2 statistic. 31 Heterogeneity between studies may be reduced through modelling relative rather than absolute differences in accelerometer outcomes between maternal educational categories. To explore this, analyses were repeated using within-study standardisation of the accelerometer outcomes. Last, in addition to controlling for the aforementioned confounders the analyses were repeated when (1) adjusting for BMI status (normal weight/overweight/obese) to assess whether the observed difference in PA and sedentary time between maternal education categories was independent of BMI status, (2) mutually adjusting for MVPA and sedentary time (ie, when MVPA was modelled as the main exposure, the analysis was adjusted for sedentary time and vice versa) to establish if differences in MVPA and sedentary time between maternal education categories were independent of each other. All analyses were conducted using Stata/SE V.11.2 in March 2014. Significance testing was two sided, and p values were used to assess the strength of evidence against the null hypotheses. RESULTS Ten ICAD studies representing 16 275 youths aged 10-18 years, met the inclusion criteria for the present analysis. Of this sample, 12 770 (78.5%) had at least 1 day of ≥500 min of valid accelerometer wear time and a self-reported measure of maternal education, and thus, formed the study population. The mean age of included participants was 12.4 years (SD=1.4) and 66.9% were female. The average number of monitored days for these participants was 5.3, with a daily monitor wear time of 790 min (SD=78). Seventy-five per cent of the participants were normal weight, 18.2% overweight and 6.8% obese. Participant characteristics by maternal grouping and study are summarised in table 1. Table 1 shows the average daily PA (CPM, minutes spent in light and MVPA), and time sedentary by study. After controlling for monitor wear, girls spent 55 more minutes sedentary compared to boys (mean: 406 min/day (SD=92.1) vs 351 min/day (SD=87.5), p<0.001). Girls accumulated 34 min less light activity (mean: 356 min/day (SD=74.7) vs 390 min/day (SD=71.5), p<0.001) and 20 min less in MVPA (mean: 22.6 min/day (SD=15.5) vs 42.1 min/day (SD=23.6), p<0.001) per day than boys. Two-way ANOVAs provided no evidence of interactions between sex and maternal education when considering their relationship with each PA outcome in turn. Consequently, data for boys and girls were combined, and models adjusted for sex. Inspection of forest plots and the I 2 statistic with 'high school' as the reference group (figure 1) indicated that moderate to high levels of heterogeneity (I 2 >25% 32 ) were present in estimates of the mean difference in PA outcomes by maternal education. Random-effects meta-analysis was used, due to these differences in the effect of maternal education across studies. Maternal education defined as high school, college and university: eight studies For all PA outcomes (CPM, sedentary, light and MVPA), and after controlling for confounders (age, season, sex and monitor wear time (where appropriate)), no significant differences were found in the combined estimate (ie, overall mean difference across studies between maternal education categories) for children of college versus high school educated mothers (see online supplementary figure S1). However, significant differences in PA were found between children of university educated mothers compared with high school educated mothers. Adjustment for sedentary time in addition to the previously mentioned confounders resulted in a higher pooled estimate of MVPA (2.2 min, p<0.001) and lower heterogeneity (I 2 =0%) in children of university versus high school educated mothers. Across all three maternal education groups, the only pooled estimates that were significant at the 0.05 level were time spent sedentary (9.5 min/day, p=0.005) and in light activity (−10.0 min/day, p<0.001) for children of university compared to high school educated mothers. For these pooled estimates there was a high level of heterogeneity within the study-specific estimates (sedentary: I 2 =76.8%; light: I 2 =71.9%), but high consistency in the point estimates either above or below the line representing 'no effect' (see online supplementary figure S1). Repeating the main analysis adjusting for BMI status as well as other confounders had no substantive impact on the results (see figure 1). Maternal education defined as primary, secondary and tertiary: two studies DISCUSSION Pooled analyses from the eight studies from the UK, Australia, Denmark, Estonia, the USA and Switzerland showed that adolescents of university educated mothers spent approximately 10 more minutes sedentary and approximately 10 less minutes in light activity per day compared with adolescents of high school educated mothers. In the two studies from Brazil and Portugal that were categorised based on primary/secondary and tertiary maternal education, pooled analyses revealed that adolescents of tertiary educated mothers were less active (approximately 40 min less light activity and approximately 7 min less MVPA/ day), and more sedentary (approximately 46 min/day), when compared with adolescents of primary/secondary educated mothers. Although 7 min more MVPA per day by children of the primary educated mothers does not appear behaviourally significant, when considering the relatively low time spent in MVPA in this sample, the difference is quite large (approximately 25% of their total MVPA average). The lack of significance in PA and/or sedentary time between children of college educated mothers and the other groups (ie, high school and university) is somewhat to be expected. There was likely little discriminatory difference in education between these neighbouring categories, as a mother could be placed into a higher category based on a short attendance at higher education. For example, a mother who completed a few weeks on a vocational course and then withdrew would still be placed in the 'college' category. Results from the present work support a small number of studies from developed countries that have used objective measurements of PA and total time sedentary to explore the association with SEP. The Gateshead Millennium Study in England found an inverse association between SEP and children's PA 22 and The Health Survey for England (HSE) showed that children grouped into the high SEP category spent approximately 17 more minutes sedentary per day. 15 Likewise, a previous analysis 23 from EYHS found a positive association between children's sedentary time and SEP in Portuguese and Estonian samples, but not in the Danish and Norwegian. The results from this study contrast evidence based on self-reported PA and/ or sedentary behaviours. Previous data suggest that children of low SEP report greater engagement in sedentary pursuits such as TV watching and other screen-based activities. 5 10 15-18 This supports the notion that self-reporting tools and objective measurement of overall PA are not measuring the same construct. In fact within the same sample there appears to be a different association between SEP and self-reporting sedentary behaviours and objectively measured sedentary time. The aforementioned paper from the HSE 15 revealed that low SEP was associated with high television watching, but lower total accelerometer assessed sedentary time. Likewise, although the association with total PA and SEP is inconsistent, 17 19 22 studies have shown that children of higher SEP participate in more self-reported organised sports. 18 33 Whereas, two separate studies of Brazilian adolescents indicated that lower SEP youth were more likely to self-report greater PA levels and, in particular, active transport. 34 35 These studies suggest that how children from different SEPs accrue sedentary time and PA may be different. Maternal education was chosen as the SEP indicator in the present pooled analysis because it was the most common indicator across the studies. The wider adoption of maternal education as an indicator of SEP by contributing studies reflects the fact that it is relatively easy to measure and garners a high response rate. Although composite measures of SEP often show stronger associations with health outcomes and behaviours, 36 conclusions based on univariate indicators, such as maternal education, may offer clearer direction for policymakers. However, although maternal education is a strong determinant of parental employment and income 37 the meaning of maternal education levels likely varies between countries, and potentially populations within a country, where education can be either compulsory or a choice. Furthermore, mothers play different cultural and societal roles depending on the country/population, while in some cultures paternal education may be more important. The data in ICAD comes from middle-high income/developed countries with generally high national levels of education. Of the 10 studies included, 8 were from developed countries and were ranked in the top 50% for tertiary education, with two (Brazil and Portugal) in the bottom 50%. 38 Association between PA and SEP may be dependent on the level of development of the country. 3 For example, in developing countries, an active lifestyle may be a necessity for those in lower SEPs, whereas in developed countries, technological advances have resulted in an erosion of lifestyle-embedded PA across all social strata, and a healthy 'active' lifestyle may require more deliberate effort. However, evidence from developing countries in rapid transition appear to support the findings from this study. For example, in rural South Africa, adolescents whose mothers had a secondary school education or higher, reported significantly more selfreported sedentary time and significantly less walking for transport than children whose mothers had no education. 39 In the older group, however, higher maternal education was associated with nearly 3 h more school and club MVPA per week. 39 Thus, future research needs to explore, across countries of varying development, the association between SEP and specific behaviours, such as active commuting, technology-driven sedentary pursuits, PE participation, sports participation, etc. Further effort is also needed to determine the impact of these behaviours on overall objectively measured PA and time spent sedentary, and how they relate to health outcomes to inform intervention and policy changes. Strengths and limitations Strengths include the large sample size and the meta-analysis of 10 individual studies, providing more robust estimates of observed associations. Physical activity and time spent sedentary were objectively measured, reducing possible misclassification, and raw data files were cleaned, processed and analysed in a standardised manner. 24 Limitations include the lack of data from low income/developing countries, the cross-sectional nature of the data and lack of consideration of important potential mediators (eg, time spent in after-school sports clubs, birth weight and birth order). 22 Our intensity threshold for MVPA (3000 CPM) was higher compared with some previous studies in youths. 40 However, when reanalysing our data using a lower threshold, the observations were materially unchanged (data not shown). Of the 10 studies included in the analyses, only one study (Project TAAG) relied on adolescent, rather than parent-reported maternal education. It is likely that adolescent-reported maternal education is less accurate than parental reported; however, validity studies have shown fair agreement between the two. 41 Some previous studies examining the association between PA/sedentary behaviour and maternal education have included a lower educational category (eg, primary school only, or 1-2 years of high school). Owing to original maternal educational coding used, this was not possible in the pooled analyses. This reduces the ability to directly compare our results with some previous studies. Last, maternal education was the sole SEP indicator considered, and it is likely that other measures of SEP have a different association with youth PA and sedentary behaviour. CONCLUSION Greater maternal education appears to be associated with lower objectively assessed PA and increased time spent sedentary in adolescents from developed countries. From a public health perspective, these results are a potential good news story as they highlight that children from low SEP may not be disadvantaged in terms of overall daily PA. Future work pooling standardised accelerometer data across countries needs to prioritise the inclusion of lower income/developing countries to fully understand the association between maternal education and PA/time spent sedentary. What is already known on this subject Socioeconomic differences in physical activity and time spent sedentary may be a mechanism linking socioeconomic position and ill health. Studies that have investigated the relationship between socioeconomic position and physical activity and sedentary behaviour have reported inconsistent findings. What this study adds In a pooled analysis of 12 770 youths across 10 studies from Europe, Australia, Brazil and the USA, those children whose mothers reported a higher level of education ( proxy for socioeconomic position (SEP)) had lower levels of objectively measured physical activity and greater time spent sedentary. This finding suggests that children from low SEP may not be disadvantaged in terms of overall daily physical activity. Future research needs to explore, across countries of varying development, the association between SEP and specific behaviours, such as active commuting, technology-driven sedentary pursuits and sports participation, and the impact these have on overall objectively measured physical activity and time spent sedentary.	2018-04-03T01:46:48.615Z	2016-01-22T00:00:00.000	{ "year": 2016, "sha1": "f44da2647f3da5e89b8c14abd169864c03cc2f60", "oa_license": "CCBY", "oa_url": "https://jech.bmj.com/content/70/6/541.full.pdf", "oa_status": "HYBRID", "pdf_src": "BMJ", "pdf_hash": "06eb929a6f9c1d5b07cfffd61fbe5a33a9307996", "s2fieldsofstudy": [ "Education", "Sociology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
59609035	pes2o/s2orc	v3-fos-license	Inhibition of inflammatory response in human keratinocytes by magnetic nanoparticles functionalized with PBP10 peptide derived from the PIP2-binding site of human plasma gelsolin Background Human plasma gelsolin (pGSN) is a multifunctional actin-binding protein involved in a variety of biological processes, including neutralization of pro-inflammatory molecules such as lipopolysaccharide (LPS) and lipoteichoic acid (LTA) and modulation of host inflammatory response. It was found that PBP10, a synthetic rhodamine B-conjugated peptide, based on the phosphoinositide-binding site of pGSN, exerts bactericidal activity against Gram-positive and Gram-negative bacteria, interacts specifically with LPS and LTA, and limits microbial-induced inflammatory effects. The therapeutic efficiency of PBP10 when immobilized on the surface of iron oxide-based magnetic nanoparticles was not evaluated, to date. Results Using the human keratinocyte cell line HaCaT stimulated by bacterially-derived LPS and LTA as an in vitro model of bacterial infection, we examined the anti-inflammatory effects of nanosystems consisting of iron oxide-based magnetic nanoparticles with aminosilane (MNP@NH2) or gold shells (MNP@Au) functionalized by a set of peptides, derived from the phosphatidylinositol 4,5-bisphosphate (PIP2)-binding site of the human plasma protein gelsolin, which also binds LPS and LTA. Our results indicate that these nanosystems can kill both Gram-positive and Gram-negative bacteria and limit the production of inflammatory mediators, including nitric oxide (NO), reactive oxygen species (ROS), and interleukin-8 (IL-8) in the response to heat-killed microbes or extracted bacterial cell wall components. The nanoparticles possess the potential to improve therapeutic efficacy and are characterized by lower toxicity and improved hemocompatibility when compared to free peptides. Atomic force microscopy (AFM) showed that these PBP10-based nanosystems prevented changes in nanomechanical properties of cells that were otherwise stimulated by LPS. Conclusions Neutralization of endotoxemia-mediated cellular effects by gelsolin-derived peptides and PBP10-containing nanosystems might be considered as potent therapeutic agents in the improved therapy of bacterial infections and microbial-induced inflammation. Background Keratinocytes, comprising 95% of the human epidermis, are the first line of defense against external harmful agents and constitute an important part of the skin's immune system [1]. In response to ultraviolet light, allergens, haptens, microbiological agents, and cytokines, keratinocytes express and release numerous immunomodulatory soluble mediators, including iNOS-derived NO, ROS, cytokines/chemokines, and prostaglandins. Those factors allow immune cells to enter the site of inflammation in the skin and contribute to the regulation of a variety of other physiological and pathological processes following an immune response [1,2]. Although (i) cytokines and other epithelial cell-derived mediators participate in maintaining normal homeostatic mechanisms in the skin and (ii) inflammation itself is an important process to fight infections and promote wound healing, dysregulation of inflammatory responses and over-activation of pro-inflammatory mediators in the inflamed skin area lead to various inflammatory diseases [3]. Therefore, down-regulation of pro-inflammatory mediators and restoration of the physiological balance between pro-and anti-inflammatory factors, is an important strategy to modulate various inflammatory skin diseases [4]. In an era of constantly increasing bacterial drug resistance, therapeutic agents that neutralize bacterial pro-inflammatory factors, including lipopolysaccharide (LPS) and lipoteichoic acid (LTA) and thereby limit the inflammatory response resulting from activation of tolllike receptors (TLRs) in host cells, might represent an innovative approach for the treatment of bacterial infections, including skin and soft-tissue infections (SSTIs) or sepsis [5,6]. Considering that SSTIs are the third most frequent cause of severe sepsis and septic shock [7], proper treatment of these infections is indispensable. Plasma gelsolin (pGSN), an isoform of a highly conserved multifunctional human protein, gelsolin (GSN), apart from its role as an element of the extracellular actin scavenger system (EASS), is a potent modulator of inflammation-mediated cellular responses with the ability to diminish inflammatory reaction of the host [8,9]. The beneficial effect of extracellular gelsolin is mediated primary via (i) selective interaction with cell wall-derived compounds i.e. LPS and LTA from Gram-negative to Gram-positive bacteria, respectively, (ii) competing with LPS-binding protein and (iii) prevention of endotoxin-mediated TLR activation [8]. pGSN binds a broad spectrum of bioactive compounds, including lysophosphatidic acid (LPA), sphingosine-1-phosphate (S1P), and platelet activating factor (PAF), enhancing the protective properties of GSN in inflammatory states [9][10][11]. Consequently, pGSN has been reported to have a beneficial effect in a variety of inflammatory-associated medical conditions, including sepsis, chronic inflammatory disorders such as chronic kidney disease, multiple sclerosis, rheumatoid arthritis, and inflammatory-mediated neurological disorders. pGSN levels have the potential to be used as a predictor of illness severity, recovery, the efficiency of treatment and/or clinical outcome [12,13]. In addition to immunomodulatory properties of the intact GSN protein, it was found that a synthetic cell membrane-permeant rhodamine B (RhB)-conjugated peptide, based on the phosphatidylinositol 4,5-bisphosphate (PIP2)-binding site of gelsolin, (GSN 160-169 [rhodamine B-QRLFQVKGRR], denoted RhB-PBP10), interacts specifically with LPS and LTA and exerts strong bacterium-killing activity against both Gram-negative and Gram-positive bacteria due to strong resemblance to natural antimicrobial peptides by displaying net positive charge, short sequence and ability to cross cell membranes [14]. Fu et al. showed that RhB-PBP10 selectively inhibits granule mobilization and secretion of oxygen radicals in FPRL1-induced neutrophils [15]. A majority of the data suggesting a beneficial function of pGSN or RhB-PBP10 has been obtained using pure extracts of endotoxins and isolated bacterial strains in a non-growing environment and studies using more complex cell culture-based experiments are still very limited. The rapid development of novel nanotechnology-based therapeutic strategies has provided new tools for treatment of infections, particularly those caused by drugresistant pathogens and has created the possibility of overcoming limitations of conventional antibiotic therapy and improve the bioavailability of bioactive substances and their antimicrobial and immunomodulatory properties. Nanoscale materials, including synthetic biodegradable polymers such as chitosan or poly-lactic-co-glycolic acid (PLGA), polysaccharide or carbon dots-based nanoparticles have been extensively studied for their promising use in biomedical technology as immunostimulants and adjuvants promoting the immunomodulatory properties of other bioactive compounds [16,17]. Among several kinds of nanomaterials, metal and metal oxide-based nanoparticles have potential use in modulation of inflammatory immune responses due to (i) limitation of inflammatory marker release, including ROS and nitric oxide in LPS-stimulated immune cells, (ii) ability to bind and remove endotoxins from the extracellular environment, (iii) modulation of gene expression and (iv) augmentation of immunomodulatory properties of other conjugated compounds [18][19][20][21]. The usefulness of nanoparticle-containing compounds was also confirmed in some in vivo models of LPS-induced pathology [22][23][24]. Our recent study of the bactericidal and immunomodulatory properties of 1,4-dihydropyridiyne (1,4-DHPs) derivatives, revealed that attachment of these bioactive compounds to aminosilane-coated iron oxide-based magnetic nanoparticles (MNP@NH 2 ) significantly improves the immunomodulatory properties of 1,4-DHPs and inhibits the proinflammatory properties of bacterial cell wall components, which is a significant advance in creation of a novel class of multimodal agents for the treatment of lifethreatening infections [25]. In this report, we characterize the immunomodulatory properties of a set of gelsolin-derived peptides based on the PIP2-binding site of gelsolin (i.e. GSN160-169) both in free form and conjugated with rhodamine B, and evaluate the impact of attachment of these bioactive peptides to aminosilane-coated and gold-decorated iron oxidebased magnetic nanoparticles on their ability to inhibit the bacterial-induced inflammatory responses in a cell culture-based model of skin infection. These studies indicate the enhancement of biocompatibility and improvement of biological features of the therapeutic peptides after their immobilization on magnetic nanocarriers when compared to their non-magnetic counterparts. We suggest that the increased anti-inflammatory properties of gelsolin-derived compounds after their immobilization on magnetic particles is related to improved cellular uptake and the properties of non-modified nanoparticles, particularly gold-decorated nanostructures to limit inflammatory response. This augmentation of peptide bioactivity suggests the possible application of MNP-based approaches in the development of improved anti-infectious therapeutic agents with combined antiinflammatory functions that diminish the excessive inflammatory reaction of the host in the response to induction by bacterial-derived compounds. Peptides The set of gelsolin-related peptides, based on the free non-conjugated sequence of GSN160-169 (QRLFQVK-GRR, or PBP10) and PBP10 conjugated to rhodamine B (denoted with RhB-as a prefix to the peptide name) with or without a terminal cysteine (RhB-PBP10-Cys and RhB-PBP10, respectively) were synthesized and provided by Lipopharm.pl (Zblewo, Poland). According to HPLC analysis provided by the manufacturer, the purity of the synthesized peptides was > 98%. Bacterial products Lipopolysaccharide from Escherichia coli O26:B6 and lipoteichoic acid from Staphylococcus aureus were purchased from Sigma Chemical Co. (St. Louis, Mo., USA). Stock solutions of LPS and LTA were prepared by suspending them in endotoxin-free water (Sigma Chemical Co.). Heat-killed S. aureus was prepared by boiling the bacteria for 7 min and then washing them three times with phosphate-buffered saline (PBS). The efficacy of the heat treatment was confirmed by culturing the bacteria overnight to ensure that there was no growth. Synthesis and physicochemical characterization of PBP10-containing nanosystems Nanosystems used in this study were obtained using iron oxide-based magnetic nanoparticles with aminosilane (MNP@NH 2 ) or gold shells (MNP@Au). The magnetic core of the nanocarriers was synthesized using a modification of the Massart method, which is based on the coprecipitation of hydrated iron chloride salts after addition of ammonium hydroxide (25%) [26]. Coreshell nanostructures with terminal propylamine groups and gold shells were obtained using Stöber and K-gold methods, respectively [27]. Following synthesis, all the nanoparticle samples were placed in an oven at 60 °C and dried into powder over 12 h. Physicochemical analysis of aminosilane-and gold-decorated nanoparticles were presented previously [27]. PBP10-containing nanosystems were obtained by non-covalent bonding including electrostatic interactions or chemisorption of thiol groups to the gold surface. Nanoparticles (MNP@NH 2 or MNP@Au) were dispersed in PBS to obtain solutions of 10 mg/mL concentration. Then nanoparticles were diluted to the concentration of 1 mg/mL and were mixed with the appropriate amount (1:1 volume ratio) of PBP peptide or PBP modified by cysteine and rhodamine (1 mg/ mL solutions in PBS buffer). Prepared solutions were incubated for 2 days (nucleation) and used for further investigations. In order to avoid the loss of agents during the preparation process, no further purification of nanosystems was performed. To evaluate the efficiency of peptide attachment, fluorescence of unbound PBP10 peptides in solutions after magnetic separation of nanosystems was measured. Fourier transform infrared spectroscopy (FTIR) spectra were recorded using a Thermo Fisher Scientific Nicolet iN10 MX FTIR microscope. For this purpose, a 10 μL sample (1 mg/mL) was dropped on the surface of a glossy metal plate, and the solvent was left to evaporate. All spectra were collected in the 4000-800/cm range by co-adding 64 scans with a resolution of 4/cm. Data analysis was performed using OMNIC software (Thermo Fisher Scientific). Hydrodynamic diameters (DLS) were measured at 25 °C using a Zetasizer NanoZS (Malvern Instruments, Ltd, UK) with integrated 4 mW He-Ne laser, λ = 633 nm. Light scattering intensity was measured at 173° in case of all samples. The concentration was 1 mg/mL of nanosystems in PBS buffer solution. The zeta-potential measurements were carried out on the same Zetasizer NanoZS analyzer using the same solutions. All measurements were carried out at 25 °C using folded capillary cells (DTS 1060). Data were generated in the form of electrophoretic mobility, which was further converted to zeta potential by application of the Smoluchowki equation. Cell culture Immortalized adult human skin keratinocytes cells (HaCaT) were grown in high-glucose Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), glutamine (2 mM/L), penicillin (50 U/mL) and streptomycin (50 µg/mL) and maintained at 37 °C in an atmosphere containing 5% CO 2 with saturated humidity. After seeding, the cells were cultured until a confluence of ~ 85% was reached. All experiments were performed in serum-free conditions. Assessment of biocompatibility of analyzed nanosystems The viability and metabolic activity of cells were measured using a microculture tetrazolium test (MTT; 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) as described previously [28]. HaCaT cells were seeded at a density of 5 × 10 4 cells/well in transparent 96-well plates and cultured until ~ 85% confluence; after that medium was replaced with serum-free medium and cells were serum starved for another 12 h. To determine the non-toxic concentrations, the peptides and their magnetic derivatives in concentrations ranging from 1 to 50 µg/mL were added to each well and incubated for 24 h at 37 °C under 5% CO 2 . The optical density at 490 nm was assessed after 2 h incubation of plates with MTT solution (5 mg/mL) and the addition of formazan salt in dimethyl sulfoxide (DMSO). The absorbance value obtained in cultures of control cells (with solvents alone) was taken as 100%. The average of all the experiments is shown as cell viability percentage in comparison to the control. Quantification of hemolytic activity The hemolytic activity of each peptide and its nanosystem was determined using a previously described method [29]. Briefly, fresh human red blood cells (RBCs) were collected and then centrifuged at 1000×g for 5 min at 4 °C. The erythrocytes were then washed three times with PBS (pH 7.2), resuspended (hematocrit ~ 5%) in PBS containing antibacterial agents at a concentration ranging from 0 to 50 µg/mL and incubated for 1 h at 37 °C. Intact erythrocytes were centrifuged at 1000×g for 5 min at 4 °C, and the supernatant was transferred to a 96-well microtiter plate. The release of hemoglobin was monitored by measuring the optical density at 570 nm (OD 570 ). 100% hemolysis was taken from samples in which 1% Triton X-100 was added to disrupt all cell membranes. Minimum hemolytic concentrations (MHC10%) are defined as the concentration that causes hemolysis of 10% of RBCs. Antibacterial testing To assess the bactericidal activity of PBP10-derived peptides and PBP10-containing nanosystems against E. coli and S. aureus bacterial strains, a killing assay based on counting colony forming units (CFU) was performed [30]. For this purpose, E. coli and S. aureus strains were grown to mid-log phase at 37 °C, re-suspended in PBS, brought to 10 8 CFU/ml (which corresponds to 0.5 OD at a wavelength 600 nm) and diluted 1:100 to obtain sample containing approx. 10 6 CFU/mL. Bacteria were then added to PBS containing different concentrations of PBP10, RhB-PBP10 and RhB-PBP10-Cys and their nanosystems at doses of 2, 5 and 10 µg/mL. After 1 h of incubation at 37 °C, the plates were transferred to ice and suspensions were diluted 10-to 1000-fold in PBS. Then, 10 μL aliquots were spotted on agar plates for overnight culture at 37 °C in order to determine CFUs. Nitric oxide assay The nitric oxide assay was performed as described previously [31] and the quantity of nitrite in the culture medium was measured as an indicator of NO production. Amounts of nitrite, a stable metabolite of NO, were measured using the Griess reagent (1% sulfanilamide and 0.1% naphthylethylenediamine dihydrochloride in 2.5% phosphoric acid). Briefly, 5 × 10 4 HaCaT cells were seeded in transparent 96-well plates and cultured until 85% confluence was reached. After that, cells were washed with PBS, and agents at final concentrations of 2, 5 and 10 µg/mL were added to each well. Simultaneously, keratinocytes were stimulated with 1 µg/mL LPS (E. coli 026:B6), 1 µg/mL LTA (from Staphylococcus aureus) or heat-inactivated suspensions of E. coli or S. aureus (~ 10 6 CFU/mL) and incubated in serum-free medium for 24 h. Next, 50 µL of cell culture medium was mixed with 50 µL of Griess reagent, incubated at room temperature for 15 min, and the absorbance at 540 nm was measured in a microplate reader. Fresh culture medium was used as a blank in every experiment. ROS formation The generation of reactive oxygen species from stimulated HaCaT cells was measured using 2′,7′-dichlorofluorescein diacetate (DFCH-DA) as the fluorescent probe. For this purpose, 5 × 10 4 of HaCaT cells, cultured until ~ 85% confluence and incubated simultaneously with peptides or their magnetic derivatives at concentrations of 2, 5 and 10 µg/mL and with 1 µg/mL LPS/LTA or heat-inactivated bacteria for 24 h. Next, cells were washed twice with PBS, and DFCH-DA in PBS at the concentration of 20 µM was added. Fluorescence was measured for 90 min immediately after the addition of the dye at excitation/emission wavelengths of 488/535 nm. IL-8 release In order to assess release of interleukin-8 (IL-8) from stimulated keratinocytes, 5 × 10 4 HaCaT cells were treated with agents at concentrations of 2, 5 and 10 µg/mL and stimulated with 1 µg/mL LPS/LTA. The medium from LPS/LTAtreated cells was harvested after 24 h of incubation and the level of IL-8 was evaluated using an IL-8 Human ELISA Kit (Thermo Fisher Scientific). Atomic force microscopy (AFM) The stiffness of cells after 24 h stimulation with 1 µg/mL LPS was measured by indentation using atomic force microscopy (Nanowizard 4 Bioscience AFM, JPK Instruments, Germany). Adenocarcinomic human alveolar basal epithelial cells (A549 ATCC ® CCL-185 ™ ) were cultured in high-glucose DMEM supplemented with 10% FBS at 37 °C with 5% CO 2 , placed on Ø 35 mm Petri dishes and allowed to attach and spread for 24 h before simultaneous stimulation with 1 µg/mL LPS (E. coli 026:B6) and treatment with 5 µg/mL of PBP10 or PBP10-based nanosystems. Immediately before the experiment, cells were placed in a CO 2 -independent buffer (Thermo Fisher Scientific, USA) to prevent changes in pH of the cellular environment during analysis. Elasticity measurements were taken with AFM working in force spectroscopy mode in liquid conditions, and cantilevers (ORC8, Bruker) with a spring constant of 0.1 N/m were used. From each tested cell, up to 64 forceindentation curves were collected in a grid of 8 × 8 pixels corresponding to a scan area of 10 × 10 µm with a maximal force of 2 nN. For indentation measurements, more than 1100 force-distance curves were recorded for each group from at least ten different cells. To determine the apparent Young's modulus of different cells, force-indentation curves were fit to the Hertz contact model. Statistical analysis The significance of differences was determined using the two-tailed Student's t test. Statistical analyses were performed using Statistica 10 (StatSoft Inc, Tulsa, OK, USA). p < 0.05 was considered to be statistically significant. Results are the average of three to six individual measurements. Synthesis and physicochemical characterization of PBP10-containing nanosystems In our study, synthesized nanosystems were obtained via physicochemical interaction between the surface of core-shell nanoparticles composed of an iron-oxide core and gold or aminosilane shells and the chemical groups of peptides. PBP10 peptides were anchored to nanoparticles structure via non-covalent interactions including electrostatic attractions or chemisorption of the thiol group to the gold surface (Fig. 1a). To confirm the presence of a gold surface on the nanoparticles, elemental analysis indicated that the gold content is ~ 2% of the nanostructure weight (Fig. 1b). To determine if PBP10 peptide derivatives were successfully immobilized on the core-shell surface fluorescent spectra were recorded. The similarity between the spectra of peptides in free and immobilized forms confirms that peptides became attached to the particles (Fig. 1c). To confirm that free, non-attached peptides do not persist in the MNP suspension, fluorescence-based measurements were performed. Results of fluorescence measurements for solutions after magnetic separation of MNP@PBP10-RhB and MNP@ PBP10-RhB-Cys are presented in Table 1. As presented, PBP10-RhB peptide is immobilized on the surface of both MNP@Au and MNP@NH2 with nearly 100% efficiency, since after magnetic separation nearly non-detectable fluorescence is recorded. The amount of unbound PBP10-RhB-Cys peptide does not exceed 5%. Considering that doses of nanosystems used in our experiments are relatively low (i.e. 1-10 µg/mL), we assume that the concentration of unbound peptide was too small to affect collected data. The nanosystems were also characterized by FT-IR spectroscopy. Figure 1d summarizes the signals, showing the absorption peaks at ~ 550/cm which correspond to Fe-O stretching. The presence of peptide molecules on the MNP surface caused the characteristic stretching vibration for silane group to overlap with signals from peptides. The absorption peaks at 1650/cm are assigned to C=O bonds. The presence of aminoacids was detected also by peaks at around 1100 and 3200/cm which correspond to C-N stretching and N-H stretching respectively. Literature data indicate that the signal from S-H stretching bonds should appear near 2500/cm, however after immobilization to the MNP surface S-S bonds might form and the signal for this group can be registered around 550/cm [32]. In our case, this signal cannot be detected, because it is merged with the Fe-O stretching signal. The zeta-potential and DLS studies were carried out for all nanosystems. Changes in the hydrodynamic diameter and zeta potential of modified nanoparticles with a comparison to bare (MNP@NH 2 , MNP@Au) were observed ( Table 2). The zeta-potential measurements have shown that the addition of RhB-PBP10 peptide to magnetic nanoparticles decorated with a gold shell causes an increase of zeta-potential from − 18.7 to − 12.3 mV. Moreover, the addition of PBP10 peptide modified by cysteine and rhodamine converts zeta-potential to a positive value (from − 18.7 to + 12.9 mV), suggesting that this peptide interacts more strongly with the surface of nanoparticles than nonmodified PBP10 peptide. Nanoparticles with aminosilane shell act similarly to nanoparticles with a gold shell. After addition of PBP10 peptide, zeta-potential increases from − 11.8 to − 7.6 mV, while the addition of PBP10 modified by cysteine and rhodamine converts the signal of zeta-potential to positive values (from − 11.8 to 15.6 mV). The observed changes in zeta-potential suggest the interaction of peptides with nanoparticles surface. A decrease in hydrodynamic diameter of peptide-modified nanoparticles is also observed. According to Table 2, the average size of gold nanoparticles decreases about 100 and 200 nm after the addition of RhB-PBP10 and PBP10 modified by cysteine and rhodamine, respectively (from 442 nm to 346 and 256 nm, respectively). Similarly, average size changes of aminosilane-based nanosystems decrease from 965 nm (bare nanoparticles) to less than 100 nm (nanoparticles decorated with PBP10 peptide modified by cysteine and rhodamine). Cytotoxicity of PBP10-containing nanosystems against human keratinocytes In the first step of evaluation of potential immunomodulatory properties of PBP10-containing nanosystems, we performed a cytotoxicity assay for human keratinocytes. For this purpose, an MTT assay based on estimation of metabolic activity was employed. According to the results presented in the Fig. 2a-c, three forms of PBP10 peptides are characterized by relatively low toxicity against HaCaT cells at doses ranging from 1 to 10 µg/mL. The lowest toxicity was noted for PBP10 and its magnetic counterparts (i.e. PBP10 + MNP@Au, PBP10 + MNP@NH 2 ) since 24 h incubation with these agents at a dose of 10 µg/mL resulted in viability of HaCaT cells in ranges from 101 ± 1.4% to 83 ± 6.9. Marginally higher cytotoxicity was noted for RhB-PBP10-Cys and its nanosystems (80 ± 6.6% for RhB-PBP10-Cys and 68 ± 0.6% for RhB-PBP10-Cys + MNP@NH 2 for a dose of 10 µg/mL). The IC 50 values for all tested gelsolin-related peptides were 98.9 ± 6.7 µg/mL, 30.3 ± 5.3 µg/ mL and 59.3 ± 11.6 µg/mL for PBP10, RhB-PBP10, and RhB-PBP10-Cys, respectively, which indicates their low toxicity against mammalian cells. Importantly, in most cases immobilization of PBP10 peptides on the surface of magnetic nanoparticles does not increase the toxic effects of the peptides. Particularly, immobilization of peptide derivatives onto gold-decorated nanosystems seems to be particularly preferable, since the IC50 values for Au-based nanosystems are higher compared to free peptides (Table 3). We also observed relatively high Results are presented as mean ± SD obtained from 3 experiments. ^I ndicates statistically significant (p ≤ 0.05) activity of PBP10-containing nanosystems comparing to non-immobilized agents biocompatibility for bare gold-and aminosilane-coated nanoparticles, considering that the number of HaCaT cells with lower metabolic activity did not exceed ~ 30% up to doses of 50 µg/mL (Fig. 2d). IC 50 values for MNP@Au and MNP@NH 2 were 74.9 ± 35.3 µg/mL and 85.0 ± 22.6 µg/mL, respectively. Additionally, up to a dose of 10 µg/mL no change in ROS generation in unstimulated HaCaT cells was noted (not shown data), which confirms that iron oxide-based nanostructures do not cause oxidative stress in our experimental settings. Based on these data, doses of 2, 5 and 10 µg/mL were chosen for further bacterial neutralization experiments. Immobilization of PBP10 peptides on the surface of magnetic nanoparticles improves their hemocompatibility In another set of experiments, we evaluated the toxicity of peptides and their magnetic derivatives using a red blood cell-based in vitro model. As demonstrated in Table 3, conjugation of GSN160-169 to rhodamine B, and supplementation of PBP10 with a terminal cysteine considerably decreased its hemolytic activity, considering that MHC10% for RhB-linked peptides rose from 6.0 ± 2.7 µg/mL to 51.1 ± 6.3 µg/mL and 72.9 ± 25.9 µg/ mL for RhB-PBP10 and RhB-PBP10-Cys, respectively. Our previous studies showed significant reduction of hemolytic activity of membrane permeabilizing agents after their immobilization on the surface of iron oxidebased magnetic nanoparticles [33], and we recorded improved hemocompatibility of PBP10-derived nanosystems, particularly when gelsolin-related peptides were incorporated into gold-based nanostructures (increase from 1.3-to 9.8-fold when compared to unimmobilized agents; Table 3). Importantly, in the great majority of combinations, at a dose of 10 µg/mL, the release of hemoglobin from RBCs did not exceed 0.5-1%. The considerable differences between aminosilane-and goldbased nanosystems suggest that the type and molecular structure of nanomaterials intended for development of nanosystems is crucial in the design of therapeutic agents with satisfactory biocompatibility. PBP10 peptides and their magnetic derivatives exert bactericidal properties against both Gram-negative and Gram-positive bacteria In our previous research aimed to evaluate the antimicrobial activities of RhB-PBP10, we reported killing activity against the Gram-negative bacteria E. coli and Pseudomonas aeruginosa and the Gram-positive bacterium Streptococcus pneumoniae [34]. With this in mind, we investigated the bactericidal activity of gelsolin-derived peptides and their magnetic derivatives against E. coli RS218 and S. aureus A1. According to results presented in Fig. 3, both PBP10 and its rhodamine B-conjugated counterparts, RhB-PBP10 and RhB-PBP10-Cys, exert significant dose-dependent bactericidal effects against both Gram-negative and Gram-positive organisms resulting in a decrease of bacterial viability by more than 95%. In the case of E. coli (Fig. 3a), the addition of a terminal cysteine to the RhB-conjugated gelsolin sequence appears to have a small inhibitory effect on the peptide's killing activity. In accordance with our previous research, the attachment of peptides to magnetic nanoparticles results in 1.7-6.4 fold increase of bactericidal activity. PBP10-based agents decrease the release of nitric oxide from stimulated keratinocytes Due to constant exposure of skin to environmental challenges, such as physical stress, trauma, chemical irritants, and infectious pathogens, keratinocytes constitute the first line of defense against external insults and participate in the innate immune response by secreting soluble inflammatory factors including NO and ROS. Therefore, inhibitors of NO and ROS production in stimulated keratinocytes might provide a valuable approach to limit inflammation [35]. With this in mind, we determined the nitrite level in the supernatants of HaCaT keratinocytes after 24 h exposure of cells to 1 µg/mL LPS, LTA or heat-killed bacterial suspensions, as an indicator of NO production, which is a well-documented response of human keratinocytes after inflammatory stimulation [35]. Intact heat-inactivated bacteria as infectious and inflammatory stimuli may better reflect a physiological encounter between keratinocytes and bacteria than the (Fig. 4b). A similar tendency was observed when results of LPS-and LTA-treated cells ( Fig. 4a, b) were compared with data obtained from samples incubated in the presence of heat-inactivated suspension of E. coli and S. aureus, respectively (Fig. 4c, d), which indicates that the activity of these agents depends strongly on the type of inflammatory stimulus. Functionalization of PBP10 using aminosilane-coated nanoparticles seems to be more efficient in augmentation of anti-inflammatory properties as measured by disruption of NO release (p-value ranging from 0.0039 to 0.0283 for PBP10 + MNP@NH 2 and RhB-PBP10-Cys + MNP@NH 2 , respectively). PBP10-based agents protect cells from oxidative stress Together with the promising results of NO release assays, significant inhibition of ROS generation in stimulated keratinocytes was noted (Fig. 5). Dysregulated ROS formation is involved in the pathogenesis of various inflammatory conditions, and agents with antioxidant properties might be considered to be potential therapeutic agents for ROS-mediated inflammatory skin diseases [36]. To study the effect of PBP10-based nanosystems on the inflammatory response of HaCaT cells, we measured ROS production after stimulation with LPS, LTA, and bacterial suspensions. According to data presented in Fig. 5, PBP10-based nanosystems strongly inhibit the inflammatory stimulus-induced ROS synthesis. Interestingly, in contrast to NO release assay, the highest activity was noted for RhB-PBP10 and its nanosystems, suggesting that intracellular localization of therapeutic agents might be crucial for effective ROS production limitation. According to the data in Fig. 5a (Fig. 5b). Most of the PBP10 derivatives, both in free and immobilized forms inhibited intracellular ROS generation when compared to untreated stimulated controls, which highlights their potential as anti-oxidative agents. The release of IL-8 from stimulated keratinocytes is decreased by PBP10-derived nanoagents The release of IL-8 is implicated in the pathogenesis of inflammatory-related diseases and is highly expressed by keratinocytes [37]. As presented in Fig. 6a, b, IL-8 levels in untreated LPS-and LTA-stimulated keratinocytes were 504.9 ± 49.39 pg/mL and 749.2 ± 55.81 pg/mL, respectively. In contrast, treatment of cells with a set of PBP10 peptides resulted in a decrease of IL-8 release to the level of 393.5-431 pg/mL in LPS-stimulated samples Treatment with PBP10 and its derivatives prevents LPS-induced changes in biomechanical properties of cells Atomic force microscopy (AFM) permits studies of cellular morphology and mechanical properties of the cell surface at the nanoscale. Considering previous reports demonstrating (i) alterations of biomechanical properties of cells in a variety of physiological and pathological processes [38] and (ii) the possibility to employ AFM-based analyses to evaluate anti-inflammatory drugs in LPS-activated cell models [38,39], we investigated changes in HaCaT cell stiffness upon exposure to LPS and subsequent treatment with PBP10 and its magnetic derivatives. We operated AFM force spectroscopy to assess cell cortical stiffness (Young's modulus) using two indentation depths: 300 nm (Fig. 7) and 1 µm (Fig. 8). According to the statistical analysis presented in Fig. 7b, the Young's modulus of the cell cortex dropped from 5.0 ± 0.2 kPa for control cells to 4.3 ± 0.07 kPa for 1 µg/mL LPS stimulated cells (p-value = 0.0176), which indicated that lung epithelial cells became softer after 24-hour LPS stimulation. The same conclusion was drawn when AFM analysis using an indentation depth of 1 µm was performed (Fig. 8b); upon exposure to bacterial endotoxin cell stiffness was decreased from 3.5 ± 0.06 kPa for control cells to 3.1 ± 0.08 kPa for LPS-treated cells (p = 0.0003). In contrast, simultaneous treatment of cells with PBP10 in free and immobilized form resulted in the reverse of the LPS effect and a partial increase in cell stiffness. A particularly visible Discussion Even during treatment with conventional antibiotics, bacteria can release endotoxins like LPS or other pathogenicity factors from their cell envelope that activate Toll-like receptors (TLRs) and induce strong inflammatory reactions [40]. Therefore, endotoxin-neutralizing agents, specifically developed to neutralize these pathogenicity factors and with potent ability to abrogate TLR-mediated inflammatory responses, represent an innovative approach for the treatment of bacterial infections. To date, some natural and synthetic agents with both antibacterial and LPS-binding and neutralizing activity, mostly from the group of cationic antimicrobial peptides (AMPs), were presented as potential candidates for treatment of microbial-related diseases, including sepsis, LPS-induced inflammation during cystic fibrosis lung disease and skin infections [6,41,42]. In this study, we employed a set of gelsolin-related peptides, derived from the PIP2-binding site of human gelsolin to develop MNP-based nanosystems able to bind and neutralize LPS and LTA. Previous studies, using transgenic gelsolinlacking mouse models, strongly indicated the beneficial effects of gelsolin and gelsolin-derived compounds in maintaining proper functions of the immune system. Witke et al. revealed that gelsolin-null mice exhibited a number of dysfunctions in inflammatory reactions, including delayed in vivo migration of neutrophils into peritoneal exudates and impaired chemotaxis [43]. Exogenous gelsolin was also found to limit neutrophil and treated simultaneously for 24 h with PBP10-based peptides and their magnetic derivatives at concentrations of 2 µg/mL (dark grey columns), 5 µg/mL (grey columns) and 10 µg/mL (light grey columns). Untreated stimulated control samples are indicated as 100%. Results are presented as mean ± SD obtained from 3 experiments. Indicates statistically significant (p ≤ 0.05) activity of tested agents compared to untreated control samples, ^i ndicates statistical significance when comparing to MNP-based agents to non-immobilized peptides ▸ migration, scavenge soluble pro-inflammatory mediators, and inhibit neutrophil adhesion to endothelial surfaces in a mouse model of acute lung injury [9,44]. Extracellular gelsolin also has a beneficial activity on macrophagemediated antimicrobial and host defense functions through a nitric oxide synthase type III (NOS3)-dependent mechanism and improved bacterial binding and killing [45,46]. These studies highlight the potential of extracellular gelsolin in the treatment of infections and microbial-associated medical conditions. Anti-inflammatory activity was also strongly suggested for gelsolinderived peptide based on the PIP2-binding sequence of gelsolin [8,9,14,34]. With this in mind, we designed peptide-based nanosystems, consisting of GSN160-169-derived peptides as bioactive compounds and magnetic nanoparticles as highly biocompatible nanocarriers (Fig. 1a). For this purpose, PBP10 peptides were attached to gold-and aminosilane-decorated nanoparticles via chemisorption of thiol groups to gold surface and electrostatic interaction with aminosilane shells, respectively, as confirmed by fluorescence analysis (Fig. 1c) and FT-IR spectroscopy (Fig. 1d). Using gold-and aminosilane-coated iron oxide-based magnetic nanoparticles as nanocarriers and compounds with additional anti-inflammatory and antimicrobial activity, we have been able to modulate the production of pro-inflammatory soluble factors, including NO, ROS, and IL-8 from cells after stimulation with purified LPS/LTA and heat-killed bacteria, E. coli and S. aureus. Several previous observations indicated that metal and metal oxide nanoparticles possess a high therapeutic value relating to its anti-inflammatory and bactericidal characteristics. Due to their interaction with cell wall components, ROS generating-properties and ability to disrupt metabolic pathways, several metal oxide-based nanoparticles have been proposed to be used in limiting microbial infections [47]. Recently, magnetic nanoparticles with aminosilane shells were successfully employed as factors modulating the anti-inflammatory activity of 1,4-DHPs [25]. Prasad et al. demonstrated the utility of nanosystems composed of long C18 acyl chains tethered to Fe 3 O 4 /Au/Fe 3 O 4 nanoflowers for simultaneous removal and detection of endotoxins from food to pharmaceutical preparations due to reversible binding with the bioactive lipid A component present on the LPS molecule [21]. Administration of gold nanoparticles (AuNPs) was found to decrease LPS-induced eye inflammatory response in a rat model of endotoxin-induced uveitis, which is determined by a decrease in TLR4 content and NF-κB activation [23]. Results from mouse models showing protection against a lethal inhalation challenge of Burkholderia mallei using a gold nanoparticle-linked glycoconjugate vaccine are also promising [24]. However, the effects of AuNP in vivo after repeated administration are still unclear and contradictory. Polyethylene glycolcoated AuNPs were shown to activate p38 MAPK/NF-κB pathways in RAW264.7 cells, which suggests potential immunotoxicity of these compounds due to their ability to stimulate macrophages to release aberrant or excessive pro-inflammatory mediators [48]. On the other hand, Ma et al. and Nishanth et al. indicated suppression of LPS-induced activation of the NF-κB signaling pathway in RAW264.7 cells and high biocompatibility of gold nanoparticles, with small inflammatory reactions noted only after prolonged exposure [19,49]. Our preliminary studies have shown that neither MNP@NH 2 nor MNP@ Au induces inflammatory effects in mammalian cells, and they do not affect NO synthesis or IL-8 release in a broad spectrum of doses and increase slightly ROS generation only at a high dose of 50 µg/mL (not shown data). In Fig. 6 Immunomodulatory properties of PBP10-containing nanosystems. The release of IL-8 from LPS-(a) and LTA-stimulated (b) HaCaT cells after treatment with PBP10 peptides in free form and immobilized on magnetic nanoparticles. Results are presented as mean ± SD obtained from 3 experiments. Indicates statistically significant (p ≤ 0.05) activity of tested agents compared to untreated control samples stimulated with LPS/LTA, ^i ndicates statistical significance when comparing MNP-based nanosystems to non-immobilized peptides contrast, we show that at doses that are reported as safe for the culture of human keratinocytes (i.e. 2-10 µg/mL) (Fig. 2), a majority of nanoparticle-based combinations exert efficient inhibition of LPS-and LTA-induced cellular effects, serving as protective agents during microbialassociated inflammation. Importantly, immobilization of PBP10 peptides on the surface of magnetic nanoparticles does not increase their toxic effects against mammalian cells (Fig. 2a-d) and provides a possibility to significantly improve the hemocompatibility of gelsolin-derived peptides (Table 3), which is in agreement with our previous research demonstrating a decrease of hemolytic activity of membrane-active agents after their immobilization on the surface of iron oxide-based nanoparticles [50]. All tested combinations are characterized by relatively low toxicity against mammalian cells, particularly in comparison with melittin, an amphipathic peptide from the group of AMPs with potent antibacterial, antiviral, and anti-inflammatory activities, proposed as a therapeutic agent with the ability to prevent Propionibacterium acnes-induced inflammatory skin diseases [51]. With regard to hemolytic activity, it is noteworthy that PBP10based agents are able to lyse RBCs, but their lytic concentrations are significantly higher than those required for bacterium-killing and immunomodulatory activity. This strongly supports the potential of these combinations in the treatment of both topical and systemic infections, as suggested before for endogenous and synthetic AMPs such as LL-37, RR, and RRIKA [52,53]. RhB-PBP10 is structurally similar to AMPs due to its cationic charge, a short sequence, ability to cross membranes and amphipathic chemical character [14]. Similar to other antimicrobial peptides, including human cathelicidinderived LL-37, RhB-PBP10 inhibits the growth of Grampositive and Gram-negative microorganisms, including Escherichia coli¸ Pseudomonas aeruginosa, Streptococcus pneumoniae and Bacillus subtilis, which results from penetration of the peptide into the membrane bilayer and Results are presented as mean ± SD. * and ^i ndicate statistically significant (p ≤ 0.05) activity of tested agents compared to untreated control samples and LPS-stimulated cells, respectively. # Indicates statistical significance when comparing MNP-based agents to the non-immobilized peptide membrane destabilization [14,34]. In agreement with these reports, we show a significant bactericidal activity of PBP10-related peptides and their magnetic nanoparticle-based nanosystems against Escherichia coli and Staphylococcus aureus (Fig. 3a, b, respectively), despite the lack of phosphoinositides (i.e. the presumed native ligand for gelsolin-related peptides) in these microorganisms. Although rhodamine B linkage to unmodified PBP10 was reported previously to be essential for the antibacterial function of RhB-PBP10 [34], we did not detect significant differences in killing activity of PBP10 and its RhB-conjugated counterpart, which is likely related to different experimental settings and different bacterial strains. More importantly, we recorded a slight decrease of antibacterial activity of PBP10, particularly against E. coli suspension, when the peptide sequence was supplemented with an additional C-terminal cysteine (RhB-PBP10-Cys) in order to improve the attachment of the peptide to gold-decorated magnetic nanoparticles ( Fig. 3a). Similar observations were noted when the activity of RhB-PBP10-Cys was tested against fungal cells and other bacterial strains (data not shown). Although a continuous and rapid development of new diagnostic and therapeutic methods, applicable in the detection and therapy of bacterial infections is observed, it is necessary to continue to search for new antimicrobial factors with high selectivity [40,54]. Despite the promising results, some evidence suggests that the employment of AMPs and their derivatives in the treatment of topical and systemic infections might be limited due to rather a poor selectivity for bacterial over mammalian cells, which results from the unspecific membrane-permeabilizing mechanism of microbial killing characterized for AMPs. We show that gelsolin-derived peptides exert a statistically significant growth inhibitory effect against clinically relevant bacterial strains, whereas cytotoxicity against human keratinocytes culture was not observed under employed conditions and concentrations. Previous Results are presented as mean ± SD. * and ^I ndicate statistically significant (p ≤ 0.05) activity of tested agents compared to untreated control samples and LPS-stimulated cells, respectively. # Indicates statistical significance when comparing MNP-based agents to the non-immobilized peptide studies suggest that higher specificity of presented gelsolin-related peptides for targeting bacteria rather than host cells is determined by the asymmetrical distribution of phospholipids in the external membranes of eukaryotic cells [14]. Another explanation of this phenomenon assumes that LPS and LTA present in external bacterial membranes are bacterial targets for PBP10 peptides and their magnetic derivatives, attracting specifically these agents and increasing peptide adsorption to the bacterial surface [55,56]. Moreover, the possibility that LPS and LTA in association with other bacterial-wall molecules form binding sites for AMPs cannot be ruled out [55]. From a variety of molecular mechanisms mediating the inflammatory responses, production of pro-inflammatory cytokines and inflammatory molecules, such as iNOS-derived NO and reactive oxygen species seems to be the most prominent. Due to a number of virulence factors, including a broad spectrum of proteins and proteases with cytotoxic activity, phenol-soluble modulins and heat-shock proteins, both S. aureus and E. coli induce strong inflammatory response in human keratinocytes due to their recognition by TLR receptors, particularly TRL-2 and TLR-4, whose expression is strongly regulated by microbial components [57][58][59]. Nitric oxide is recognized as an important cellular modulator and signaling molecule, produced by a variety of human cells, including keratinocytes and macrophages during inflammatory reactions in the skin, where NO mediates cytotoxicity, controls bacterial infection and acts as an immunoregulatory factor [35]. As NO is synthetized primarily by constitutively expressed isoforms of NO synthases present in skin, a significant amount of keratinocyte-derived NO comes from the inducible isoform of this enzyme (iNOS), whose expression is increased after challenge with bacterial endotoxins, cytokines (e.g. IL-1β, IFN-γ, TNF-α, IL-8) and neuropeptides [60]. Importantly, overexpression of iNOS and the resulting augmentation of NO generation was demonstrated in some inflammatory-based and autoimmune-related skin diseases, such as psoriasis, sunburn erythema and edema [61]. Despite the fact that a compelling number of studies demonstrated that NO generation by skin cells is a crucial component of innate immunity serving as protective agent against several pathogens including Mycobacterium leprae, Leishmania spp., E. coli, and Candida albicans, excessive production of nitric oxide can lead to edema, prolonged inflammation and injury by promoting the infiltration of macrophages and lymphocytes into the tissue [35]. In effect, recognition of the pleiotropic biological activity of NO in a variety of medical conditions has resulted in the development of therapeutic approaches aimed to modulate NO production and a few pharmacological agents that both release NO and limit its production have been described [35,62,63]. It was confirmed that inhibition of the inducible NO pathway in human keratinocytes is partially involved in therapeutic effects of retinoic acid derivatives [64] and was also presented to be valuable to relieve symptoms of flushing and erythema in the skin [61]. In addition to reports demonstrating the anti-inflammatory features of agents limiting the NO generation in some medical conditions, antioxidants have been proposed to be beneficial in oxidative stress-induced inflammation. To date, reactive oxygen species released from inflammatory cells into the extracellular compartments and causing local propagation of the inflammatory reaction and tissue damage are recognized as one of the critical factors in inflammatory skin diseases [65]. Importantly, reactive oxygen intermediates interact with NO to generate a second line of reactive molecules that can attack a number of nucleophilic extracellular and intracellular targets. Previously, superoxide anions have been reported to react with NO to form peroxynitrite, a potent inducer of lipid and protein peroxidation, tyrosine nitration and other free radical-mediated reactions in Propionibacterium acnes dependent skin infection [66]. Similarly, keratinocytes exposed to UVB or arsenite produce both superoxide anions and NO, potentially leading to the peroxinitrite formation and thus DNA and protein damage in keratinocyte cultures [67]. Considering the implication of TLR2-mediated pathway in P. acnes-induced acne and reports demonstrating the significance of this receptor in staphylococcal infections, we hypothesize that similar ROS-and peroxynitrite-associated mechanism can mediate the apoptosis of bacterialstimulated keratinocytes in our experimental model (data not shown). Therefore, the inhibition of ROS production or scavenging the released ROS may be important in preventing excess tissue damage during skin infection and inflammation [65]. With this in mind, we show that upon treatment of endotoxin-and bacterial-stimulated keratinocytes with PBP10-based nanosystems both NO (Fig. 4) and ROS (Fig. 5) release are significantly reduced. It cannot be ruled out that this limited effect is partially conditioned by keratinocyte death, but since ROS and NO production was inhibited to a significantly stronger degree, than results from cytotoxicity measurements (for instance to ~ 20% for RhB-PBP10 + MNP@Au when cytotoxicity did not excess 30% depending on the tested agent) we suggest that this effect has to be conditioned by immunomodulatory properties of the nanosystems. This effect seems to be stimuli-dependent. We hypothesize that the activities of nanosystems between LPSand LTA-treated samples are determined by differences in binding, and thus neutralization of these pathogenic factors by gelsolin-derived nanosystems. Additionally, despite the fact that the tendency in the detected neutralizing activity of PBP10-containing nanosystems does not significantly fluctuate between cells treated with purified LPS/LTA and heat-inactivated bacteria, we observed some alternations in biological activity in bacterial versus extract-treated samples. We suggest that slight differences between these pairs might be conditioned by additional features, including bactericidal and membrane-permeabilizing properties and other bacterium-associated factors, such as a wider spectrum of microbial-derived stimulatory compounds present on the surface of microorganisms, which are able to stimulate TLRs and thus, affect the generation of these soluble mediators in HaCaT cells and induce inflammatory responses [59]. This issue needs to be thoroughly investigated. In correlation with results presenting a statistically significant suppression of NO and ROS generation, a decrease in the amount of IL-8 detected in the extracellular environment was observed (Fig. 6). Because psoriatic keratinocytes have previously been shown to over-express mRNA transcripts for iNOS in the response to IL-8 [68] and cultured keratinocytes have been reported to produce a large amount of IL-8 during TLR2-mediated P. acnes skin infection [66], the detection of a simultaneous decrease of all inflammatory-associated mediators, i.e. nitric oxide, ROS, and IL-8 seems to be reasonable. Moreover, similarly to the results of NO generation assay (Fig. 4) we observed stimuli-dependent effects of PBP10containing agents-PBP10 and RhB-PBP10 and their nanosystems seemed to be most effective in the limitation of IL-8 release in LPS-and LTA-treated cells, respectively (Fig. 6a, b), which highlights the link between nitric oxide and IL-8 generation and release. One of the newer analytic approaches aimed to evaluate the anti-inflammatory drugs in LPS-activated cell culture-based models is atomic force microscopy, a high resolution imaging technique permitting to observe and analyze biological samples under physiological conditions using nanoscale resolution at the single cell level [69]. Recent studies indicate that changes in nanomechanical properties of cells reflect the transformation of cell physiology during the processes of molecule presentation and recognition, signal sensing, and increased expression of surface molecules or activation of cells. The scale of this phenomenon might be quantitatively described using the relative value of the Young's modulus [70]. Importantly, the reverse of stimuli-induced cellular effects might be recognized as an indicator of biological activity of some drugs [39]. Previously, AFM was successfully used as an analytic tool to detect LPSinduced inflammatory responses in macrophages and monocytic cell cultures [38,39], to analyze mechanisms of pathophysiological neutrophil mechanics in endotoxemia-related inflammatory conditions [71] and to assess the reduction of stiffness-dependent exacerbation of inflammatory processes [72,73]. In our study, we used AFM to investigate whether a set of PBP10 peptides and their magnetic derivatives possess the potential to inhibit LPS-mediated alterations in nanomechanical properties of endothelial cells, and we found that endotoxininduced decrease of cellular stiffness might be effectively turned back using PBP10 and its magnetic counterparts. To better understand the mechanism of PBP10-mediated effects we performed AFM elasticity measurements using two indentation depths: 300 nm and 1 µm, which allowed to assess local and average effect for the whole analyzed cell, respectively. In accordance with previous studies indicating the non-homogeneity of the cytoskeleton and pointing out the differences between the mechanical properties of cell's surface and deeper parts of a cell [69], we recorded higher Young's modulus values for smaller indentation depths, which corresponds to stiffness of regions rich in the network of actin filaments (Fig. 7b) and lower relative values of this parameter, when overall stiffness of the cell was analyzed (Fig. 8b) [69]. Importantly, we observed statistically significant changes in Young's modulus of LPS-treated cells, which indicates alterations in actin reorganization upon endotoxin stimulation [39]. Moreover, incubation of LPS-treated cells with PBP10 peptide and MNP@Au/MNP@NH 2 resulted in an increase of cellular stiffness to values comparable with non-stimulated control samples, which is most likely related with LPS binding and neutralizing properties of PBP10 peptide and nanoparticles. Pi et al. in their research aimed to evaluate the anti-inflammatory activity of dexamethasone and quercetin in a macrophage-based in vitro model and demonstrated that anti-LPS effects of these compounds are determined by the decreased binding of LPS and CD14 receptor on the surface of RAW264.7 [39]. We hypothesized that a similar phenomenon might occur in our experimental settings, nevertheless more complex single cell-based analyses are required to state this. An interesting observation noted during this study was a statistically significant PBP10-mediated increase of cellular stiffness when large indentation depth analysis was performed, which indicates that treatment with this peptide raises not the only stiffness of cell membrane, but also the whole cell interior (Fig. 8b, c). Based on our previous report, indicating an increase in stiffness and F-actin content in the cortical region of lung epithelial cells upon treatment with LL-37 peptide and accompanied by a decrease in transepithelial permeability and Pseudomonas aeruginosa uptake [73], we suggest that stiffening of cells in the response to treatment with PBP10-containing agents, both in free and immobilized form might provide an additional protective mechanism of these agents in microbial-associated medical conditions. Conclusions The experiments described here demonstrate that PBP10, PBP10-derived peptides and nanosystems consisting of iron oxide-based magnetic nanoparticles coupled to these compounds might be useful in the neutralization of bacterial pathogenicity factors, including LPS and LTA and thus, can be successfully employed in the modern therapy of microbial-associated conditions, including drug resistant infections and bacteria-induced inflammatory states. Importantly, the attachment LPS/ LTA-binding peptides to the surface of MNPs obtained via electrostatic interaction/chemisorption augments the anti-inflammatory and bactericidal capability of gelsolin-derived peptides with subsequent improvement of their cyto-and hemocompatibility, which highlights the potential of nanotechnology-based approaches. Since the severity of inflammatory states is conditioned by the balance between inflammatory and anti-inflammatory factors, inhibiting the release of nitric oxide, ROS, and IL-8 and reversing endotoxin-induced cellular effects using appropriate antioxidants and anti-inflammatory molecules could be considered as a potential treatment of SSTIs and sepsis.	2019-02-04T13:48:22.752Z	2019-02-02T00:00:00.000	{ "year": 2019, "sha1": "1d09793b17b4c69dbc6317520c047f488bf4a264", "oa_license": "CCBY", "oa_url": "https://jnanobiotechnology.biomedcentral.com/track/pdf/10.1186/s12951-019-0455-5", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "1d09793b17b4c69dbc6317520c047f488bf4a264", "s2fieldsofstudy": [ "Chemistry", "Materials Science", "Medicine" ], "extfieldsofstudy": [ "Medicine", "Chemistry" ] }
247579795	pes2o/s2orc	v3-fos-license	The Evaluation Method of English Teaching Ability in View of Internet of Things This paper firstly investigates the current situation of English teaching evaluation research at home and abroad and the evaluation of basic English teaching through literature review. Most of the research involves personal experience summary, lack of academic, and empirical research proof. Research starting points are often very similar, with repetitive content. Most of these studies are limited to teaching evaluation or learning specific English knowledge, and the evaluation of teaching methods is limited. The article explains the importance and scope of English teaching evaluation research through the Internet of Things technology and explains that the guiding theories of this English teaching evaluation research are system theory, structure theory, and multiscientific theory. Finally, this paper conducts relevant experimental research and analyzes all aspects of English teaching. Through experiments, this paper finds that students’ input and level in collective or group English activities are both above 4.2 points. The self-assessment method obtained a score of 4.8 or more in the teacher’s evaluation. It is necessary and realistic to establish an independent teacher-student-centered teacher education evaluation system, to optimize the teacher’s beliefs of English teachers, to enrich and update the knowledge system of English teachers, to develop the practical wisdom of English teachers, to focus on students as the center of evaluation and decision-making, and to develop good teaching evaluation and decision-making habits. Introduction Although English is a language and a communication tool, it conveys the information of society's politics, economy, culture, and other aspects, and itself has become an important part of a country's comprehensive national strength. The need for "high-quality talents with international communication skills and international competitiveness" further emphasizes the importance of English teaching. In recent years, China's English teaching has made great progress, but it still cannot meet the market's demand for high-level talents, including high-quality and skilled talents. The survey shows that, on the one hand, social transformation and upgrading require a large number of skilled talents; on the other hand, the technical talents cultivated by colleges and universities cannot meet the social needs. However, for a long time, the theoretical research and practical exploration of English teaching evaluation have been in a marginal position, and the traditional teaching evaluation method is poor, which increasingly hinders the progress of English teaching. As an important part of teaching, teaching evaluation is an effective way and means to solve the existing problems in teaching. However, the current theoretical research and teaching practice of English teaching evaluation are far from each other. The research on English teaching evaluation has certain theoretical and practical significance. To a certain extent, it can be said that there has been educational evaluation since the existence of human education. China's educational evaluation has a long history, but it is the West represented by the United States that has elevated educational evaluation to theoretical research and has achieved great results. Although China has made some theoretical research on educational evaluation in recent decades, it mainly draws on Western educational evaluation theories. There is no systematic theoretical analysis and practical investigation. Most of the researches combined with specific disciplines are aimed at novice or expert teachers of mathematics and chemistry in junior and senior high schools, and there is no research on outstanding English teachers in colleges and universities. Therefore, studying the teaching evaluation decisions of outstanding English teachers in colleges and universities has both theoretical and practical significance. Therefore, to carry out the research on the evaluation of English teaching will help enrich people's rational understanding of the evaluation of English teaching, broaden the research perspective of the evaluation of English teaching, and make up for the lack of theoretical guidance of the evaluation of English teaching. Carrying out the research on the evaluation of English teaching will help the evaluation subject and the evaluation object to deeply understand the problems existing in the evaluation of English teaching. It urges evaluation subjects and evaluation objects to change evaluation concepts, improve teaching methods or learning methods, adjust teaching strategies or learning strategies, improve teaching efficiency or learning efficiency, and create a good English teaching evaluation environment. Therefore, it can improve the scientific, artistic, and fairness of teaching evaluation, improve the quality and efficiency of teaching evaluation, and then improve the quality of English teaching as a whole. Related Work Experts at home and abroad have also conducted a lot of research on the evaluation of English teaching ability and the Internet of Things. Villanueva et al. believe that teaching evaluation is an important aspect of higher education. There is a lot of information based on the best existing teaching evaluation methods, but there is a lack of research on teaching evaluation methods, especially for engineering projects [1]. Dehon et al. provide educators with an assessment tool with systematic evaluation and development capabilities that can provide educators with real-time operational feedback and support improved instructional feedback [2]. Razzaque et al. believe that the characteristics of IoT, including hyperscale IoT, heterogeneity at the device and network level, and the large number of spontaneous events generated by these objects, will enable the development of various applications and services. This is a very challenging task [3]. Lin et al. provide a combination of fuzzy/edge computing and IoT that enables the deployment of network edge computing service devices. The purpose is to improve user experience and service stability in the event of a failure. With the advantages of distributed architecture and proximity to end users, fuzzy/ edge computing can provide faster response and higher quality services for IoT applications [4]. To reduce the amount of data collected on IoT, Xue et al. increase the speed of processing big data. To reduce the data collected on IoT, he proposed a compressed sensing sampling method [5]. Kshetri examines the role of blockchain in strengthening the security of the Internet of Things (IoT). It contains key underlying mechanisms involving the blockchain-IoT security relationship [6]. However, these studies pointed out that the earlier education evaluation covered a narrower scope and the evaluation method was more monotonous. Modern education evaluation covers a relatively wide range, and the evaluation method is more flexible, so these views are not recognized by the public. Evaluation System Design of the Internet of Things The design of the system includes the overall program structure, network structure, database structure, and program design of each functional module. Through the design of the system, the business logic relationship is more clearly described, and the design is described in detail. After analysis, we determine that the Web-based teaching evaluation and analysis system adopt the structure of B/S mode. The system provides users with a personalized interface according to the system user's request, user category, and access authority, which is convenient for user operation and maintenance. The database layer is responsible for the data storage of the system. The database management system used in the teaching evaluation management system based on Extjs technology is SQLSERVER2005, which ensures the storage of data [7]. The system can well meet the requirements of integration with other systems by using the framework integration technology. The specific architecture of the system is shown in Figure 1 [8]. According to the analysis of the business function requirements of the system in the previous development, it can be seen that the Web-based teaching evaluation and analysis system mainly include the following functions: management of student letters, teacher information management, student evaluation, teacher mutual evaluation management, branch leadership evaluation management, comprehensive evaluation management, teaching evaluation query management report management, system user management, and system data management. Its functional structure diagram is shown in Figure 2 [9]. After completing the collection of these information, the collected data can be correlated, and the data information can also be modified and queried. The main function of the three subsystems of student, teacher, and leader evaluation management is to complete the evaluation of the quality of the corresponding teachers' teaching work. After the collection of teaching evaluation data of each object in pairs, the evaluation integrated management subsystem will manage and control it in the evaluation integrated management module. It can be seen from Figure 2 that the system is mainly composed of ten subsystems. Among them, the main function of the student information management subsystem is to view course information and personal student status information, modify personal student status information, and change password [10]. The main function of the course information management subsystem is the collection of course information, such as the name of the course offered by the school this semester, the teacher, the time, place, and class of the course. After completing the collection of these information, the collected data can be correlated, and the data information can also be modified and queried. The main function of the three subsystems of student, teacher, and leader evaluation management is to complete the evaluation of the quality of the corresponding 3 Wireless Communications and Mobile Computing teachers' teaching work. After the comprehensive evaluation management subsystem mainly completes the collection of teaching evaluation data for each object, it will be managed and controlled in the evaluation integrated management module. By collecting the data into the database, it organizes and manages the data, performs statistical operations, and generates reports and a series of processing operations related to the collected data. It is convenient for users to access and manage teaching work [11]. The main function of the teaching evaluation query management subsystem is to allow users to easily browse the teaching evaluation results. The main function of the system user management subsystem is to manage system user roles and set permissions for roles. The system data management subsystem is mainly to backup and restore the basic data in the system, as well as the evaluation situation and results. According to the demand analysis of the system, determine the overall workflow activity diagram of the system, as shown in Figure 3. Student, Teacher, and Leader Evaluation Management Subsystem. Instructional assessments are evaluated by three system users: students, teachers, and leaders. When they log in to the system, the system displays the teaching evaluation items and the evaluators according to different roles and permissions, and the system users evaluate the evaluators. Finally, after the evaluation is over, the collected teaching evaluation data information is stored in the system database, and the specific teaching evaluation management flow diagram is shown in Figure 4 [12]. Database Design. In the development of software projects, a database is often used to save the basic information Wireless Communications and Mobile Computing and related data of the system. There are two basic information tables; one is an independent table that does not need to depend on any table, such as a department table and a course type table. The quality of a database design also directly determines the success or failure of the system. The design of the database must be made on the basis of fully analyzing and understanding the data requirements. According to the article, the class diagram of each model module of the system is converted into a data model, and the class diagram of each module in the management system is correspondingly converted into a data table, and the relationship of each table is shown in Figure 5 [13]. The form design in the system mainly includes various information tables, evaluation type information table, which is used to store evaluation type code, evaluation type name, and evaluation type description, as shown in Table 1 [14]. Class is a data table used to store class information opened by the school. It mainly includes class code, class name, class teacher, department, major, grade, description, remarks, etc. Among them, the class number is the primary key, and the class name is required to be input by the user as the analysis and management of teaching evaluation in the table design. Except for the class number and class name, other information in the table is only used as a basis for identification and does not require certain input of operation data. The specific class storage information table is shown in Table 2 [15]. The teaching evaluation system for college teachers implemented in this paper uses ASP for programming. It Figure 4: Student, teacher, and leader assessment management flowchart. Wireless Communications and Mobile Computing provides a set of solutions for the complete evaluation of teachers' classroom teaching, according to certain evaluation process and authority setting, which comprehensively evaluates teachers in four ways: student evaluation, teacher mutual evaluation, expert supervisors, and system administrators. It calculates the teacher's final score according to the weight and generates an evaluation report. And it can comprehensively analyze the evaluation results of teachers in the whole school through horizontal comparison and historical data analysis [16]. The overall flow chart of the system is shown in Figure 6. Throughput Performance Analysis of Candidate Preamble Allocation Strategies. We will model and analyze the throughput of candidate preamble allocation strategies and the proposed strategy. Let us name each strategy first, the no-overlap separation strategy is named NOSS (NoOver-lapSeparatedStrategy), the one-size-fits-all intersecting strategy is named OSS (OverlapSeparatedStrategy), and the scalable preamble allocation strategy with reservation is named SRS (ScalableReservedStrategy) [17]. In the NOSS strategy, the RACH throughput of HTC and MTC services can be expressed as: In the NOSS strategy, high-priority and low-priority services in any area include HTC services and MTC services. Therefore, the RACH throughput of high-priority traffic Under the OSS strategy, not all HTC services will try to use the preamble in region 1. Among them, only ðN − xÞ/ N100ð%Þ random access attempts fall into area 1, and the other random access attempts fall into area 2. Therefore, the packet arrival rates in area 1 and area 2 can be obtained as: So the RACH throughput of HTC and MTC services can be expressed as: The article assumes that in the OSS strategy, the high and low priority services of HTC services in area 1 are proportional to try the preamble in area 2, so the RACH throughput of high-priority services and low-priority services can be expressed as: Its definition is shown in Formula (9), and there is obviously the following mathematical relationship: By combining Formulas (2), (7), and (9), Formula (10) can be obtained. Figure 6: Overall flow chart of teaching evaluation system. Wireless Communications and Mobile Computing This numerical result is the number of random access preambles for region 2 in the OSS policy. The RACH throughput size of HTC service and the RACH throughput size of high-priority service are as follows. As of March 20, 2021, the "ProQuestResearchLibrary" database has searched "teachingdecision-making" and obtained 408 search records, the number distribution of which is shown in the figure. It can be seen that foreign research on teachers' teaching decision-making shows an upward trend. To a certain extent, this shows that in recent years, the research on teachers' teaching decision-making has received more and more attention and attention from researchers [18]. Figure 7 shows the number distribution of foreign teaching decision-making research. English Teaching Ability Evaluation Methods The article uses CNKI, JSTOR, Wiley Online, ProQuest, and other databases to collect domestic and foreign literature on teacher decision-making and teacher teaching evaluation decision-making, so that the reference of the experiment is reasonable and wellfounded. By sorting out, analyzing, and summarizing the literature, the author can more clearly understand the current research progress and results of teachers' teaching decisionmaking and teaching evaluation decision-making. It lays a solid theoretical foundation for the research and also inspires the author to find a new research entry point. It makes some supplements on the theoretical basis of predecessors [19]. 4.2.2. Questionnaire Survey Method. The article divides the questionnaire into two parts: (1) the four aspects of teachers' age, teaching age, and educational background and (2) the main body of the questionnaire has a total of 15 questions, and the designed topics include the focus of English teachers' teaching evaluation decisions, content, teaching evaluation, and decision-making methods and subjects. In order to find out the possible problems in the questionnaire, the author conducted a small-scale pilot survey on 5 teachers before starting the formal survey. According to the feedback information obtained, the author made appropriate modifications and additions to some questions in the questionnaire and finally determined the questionnaire to be issued; 30 questionnaires were distributed. 30 copies were recovered, the recovery rate was 100%, and the effective rate was 100%, which met the statistical requirements, and the data could be used for research. In addition, the data collected from the questionnaire is mainly used for descriptive analysis using the SPSS 19.0 statistical software. And by calculating its mean and standard deviation, the evaluation content, method, subject, and influencing factors that teachers focus on in teaching evaluation decision-making are comprehensively analyzed. Interview Method. 5 excellent English teachers with different background characteristics were selected from 30 research subjects for interviews (Table 4 shows details). The interviews with the case teachers are mainly to ask questions about some classroom phenomena discovered by the author while listening to the class, for example, asking questions about a specific teaching link and question in the classroom or listening to teachers' evaluation of their own teaching and the adjustment plan adopted [20]. Observation Method. The author selected three teachers X, Y, and Z from the 5 teachers interviewed to conduct classroom observation and participate in their entire teaching process. A comprehensive investigation was conducted through teachers' behavioral performance, interview records, and questionnaires. The Content of Teaching Ability Evaluation Decision- Making. The article is mainly aimed at excellent English teachers in colleges and universities; most of the teachers pay more attention to the performance of students. In the questionnaire survey, the author designed 6 questions about the evaluation content from the perspective of students: Content 1. The evaluation content is based on the degree and level of interaction between students and teachers in English. Content 2. The evaluation content is based on the students' input and level in group or group English activities. Content 3. The evaluation content is based on the development of students' English knowledge, listening, speaking, reading, and writing skills and affection. Content 4. The evaluation content is based on whether students will apply the English knowledge they have learned in their daily life. Content 5. The evaluation content is based on whether students have mastered foreign culture and whether they are interested. Content 6. The evaluation content is based on whether students can think in English. After each teaching, teachers must evaluate students' learning and their own teaching. There are various evaluation methods, and how will excellent English teachers make evaluation decisions. What is the difference between the way of its choice and the way of ordinary teachers? In the questionnaire survey, six questions about evaluation methods were designed: Method 1. English test is the main channel for obtaining information related to evaluation. Method 2. Understand students' English learning and progress through activity observation, portfolio, and behavioral evaluation. Method 3. In addition to oral assessments, written or nonverbal assessments will be used, such as questionnaires, reports, and role-playing. Method 4. Evaluation feedback not only pays attention to students' answers but also pays more attention to the process of students learning English. Method 5. Teaching evaluation will be made in class according to the actual teaching progress and situation. Method 6. After each class, they will evaluate their own teaching and reflect in time and make positive adjustments to the original teaching plan. By analyzing the data in Figure 8, it can be concluded that for the six evaluation contents, "taking the development of students' English knowledge, listening, speaking, reading and writing as the evaluation content (content 3)" is the most valued by excellent English teachers in colleges and universities. The average value is above 4.4. This shows that excellent English teachers in colleges and universities do not focus on students' test scores, but on students' various gains, experiences, and development of abilities in the process of English learning. Excellent English teachers in colleges and universities choose to evaluate students' English learning from a long-term perspective and are responsible for students' future. The second is "to evaluate the content (content 2) of students' input and level in collective or group English activities", with an average value of more than 4.2. This shows that excellent English teachers in colleges attach great importance to students' classroom participation, because students have limited time to speak English. Except for the 45 minutes in class, they rarely have the opportunity and time to speak English after class. Therefore, most excellent English teachers in colleges and universities choose to take students' investment and level in English activities as an important evaluation content. Next is "whether students will use the English knowledge they have learned in their life as the evaluation content (content 4)" (mean value 3.8), which the author did not expect. Maybe it is because college English teaching is no longer just grammar indoctrination and exam-oriented education, but more importantly, let students learn how to use authentic English and ultimately achieve students' individual development and barrier-free communication. The option that is not valued by teachers is "the evaluation content (content 6) based on whether students can think in English" (mean value 2.1). This is expected, because all the students in the class are Chinese, and they have been influenced by Chinese traditional culture since childhood, forming a certain Chinese-style thinking mode. It is too difficult to change a certain deep-rooted thinking mode, so excellent teachers do not take this content as an important reference for evaluation and decisionmaking. The biggest gap is in the item "whether students have mastered foreign culture and whether they are interested in the evaluation content (Content 5)." It shows that on this issue, teachers' teaching beliefs may be very different. Some excellent English teachers attach great importance to students' absorption and mastery of foreign cultures and believe that to learn a language well, one must master its culture. Only in this way can a language be internalized into its own thing in the true sense. And only with interest in foreign cultures, curiosity, intellectual curiosity, potential language learning ability, and even talent will be stimulated. Only when students want to learn English from their heart can they learn English easily; instead of dealing with English exams, they regard English learning as a pain and a burden. The standard deviation of "taking the development of students' English knowledge, listening, speaking, reading, writing and affection as the evaluation content (content 3)" is only 0.11, which is the smallest difference among all options. In other words, excellent English teachers pay attention to the English knowledge, listening, speaking, reading, writing, and emotional development that students have learned as the primary content of their evaluation and decisionmaking. It comprehensively evaluates students from multiple aspects and angles and strives to make a more reasonable and comprehensive evaluation of students, rather than just focusing on students' academic performance. In this way, some students who are not very strong in written test ability but have rich knowledge of foreign cultural background or have better oral expression ability are greatly encouraged. Let these students also have a successful learning experience, become more interested in English learning, and will not be frustrated or disgusted with English learning due to lower written test scores. Because excellent teachers make them understand that college English learning is no longer for exams, but for students' individual needs and future development, the focus is on students' ability to master and comprehensively use English [21]. The Main Body of Teachers' Teaching Evaluation Decision-Making. When evaluating students' learning and teachers' teaching, teachers are the most common and most important subject of evaluation, but the role of others in the evaluation process cannot be underestimated. Then, whether excellent English teachers will be different when they make the main decision-making of teaching evaluation. In the questionnaire, two options about the evaluation subject are designed: (1) encourage students to self-evaluate, group mutual evaluation subject and (2) participate in research group activities regularly and listen to the opinions of experts and colleagues, and descriptive analysis of the main body of teachers' evaluation and decision-making, please refer to Figure 9 for details. By analyzing the data in Figure 9, it can be concluded that "encouraging students' self-evaluation, group self-evaluation, students' mutual evaluation, group mutual evaluation, etc. (subject 1)" is the most popular option for excellent teachers, with an average value of over 4.8. It shows that excellent English teachers do not only take themselves as the main body of evaluation decision-making when making evaluation decisions. Instead, it allows students to fully participate in the evaluation decision-making, giving them more voice and initiative. Allowing students to self-evaluate, on the one hand, fully gives students the opportunity to speak English and on the other hand helps to reduce the embarrassment of some introverted students who are afraid of teachers evaluating themselves in public. Because many students will be afraid to speak English again due to one or several class evaluations by teachers, they always feel that they will make mistakes when they speak and will be laughed at by other classmates. The peer evaluation of groups and students will make students more relaxed, not as stressful as teachers evaluate students, and students can more easily see the gap between themselves and others, their own group, and other groups. Everyone learns from each other, learns from each other, and gradually forms a healthy competition. Ultimately, it promotes students' individual and group English learning and at the same time enhances the team strength between groups, cultivates students' team spirit, and invisibly cultivates students' good quality. This is very helpful for the social ability of students to enter the society in the future. The average value of the option "Frequently participate in research group activities and listen to the opinions of experts and colleagues (subject 2)" also reached 4.3 or more. It shows that when teachers make evaluation decisions, they do not only focus on themselves, but actively participate in the school's research group activities. They patiently listen to and share the teaching experience of others and reflect on the difference between their own teaching and others' teaching and collide with the spark of inspiration in teaching. They summarize how the same teaching content can be more effective. What are the highlights of other teachers' classes, whether they can be transplanted into their own teaching, and whether there is anything they can improve. At the same time, we humbly accept the valuable opinions and suggestions of other experts, teachers, and colleagues, so as to make teaching plan more perfect and teaching more comfortable [22]. As the direct object of teachers' teaching, students naturally play a role that cannot be underestimated in teachers' teaching evaluation and decision-making. Whether it is an individual student or a group, it affects the orientation and effect of teachers' teaching decision-making all the time, as well as their decision-making on teachers' teaching evaluation. Only by studying and understanding students can teachers make reasonable and effective teaching evaluation decisions. Teachers should focus on understanding the interests, strengths, personalities, learning styles, growth experiences, and family circumstances of individual students. Through the research of three case teachers, it can be found that they are well aware of the characteristics and situations of most of the classmates in the class. As Teacher Y said, textbook knowledge is dead, but students are alive. Therefore, teachers must pay attention to observe and summarize the characteristics of each student in the class. The teaching environment is also the main factor considered in the teaching evaluation decision of excellent English teachers, mainly including the school cultural environment and the classroom teaching environment. The data is shown in Figure 10 [23]. The culture of equality, innovation, and individuality formed over the years in Chongqing University, where the case teachers are located, subtly affects the teachers' attitudes and methods to solve problems. When 11 Wireless Communications and Mobile Computing talking to the case teacher about the school, Teacher Z said, "One of the things I like most about the school is that there are fewer constraints and more autonomy is given to the teachers. We can make own evaluation criteria and try different teaching methods, why not do it?" Teacher Y also felt the same way, "Schools and colleges provide teachers with many opportunities for training, studying abroad, and sometimes conducting teaching and research activities. Through the exchange of ideas, sparks will collide, which is conducive to promoting teachers' reflection and improving their own and others' teaching. To make more accurate evaluations, so as to improve their own evaluation and decision-making ability and teaching quality." The article argues that no matter whether the case teacher is aware of it or not, the school culture actually exists in the teacher's teaching behavior consciousness in a potential way. This makes case teachers consciously converge to the cultural atmosphere in which they live when making evaluation decisions. At the same time, the positive teaching and research culture enable case teachers to always humbly examine their own teaching ideas, content, and methods. As the case teachers mentioned, because of their participation in teaching and research activities, they developed the habit of carrying out teaching reflection and self-evaluation after class, which made the case teachers successfully surpass themselves again and again. Discussion After data analysis, it is found that the most important content of outstanding English teachers in colleges and universities when making teaching evaluation decisions is "the development of students' English knowledge, listening, speaking, reading and writing skills and affection". The second is "students' level of investment in group or group English activities", and the least valued by teachers is "whether students can think in English". In terms of evaluation methods, English tests are still the main channel to obtain information related to evaluation, but excellent English teachers will also use written or nonverbal evaluations, such as anonymous questionnaires and reports. As far as the evaluation subject is concerned, in addition to the teachers themselves, excellent English teachers endow students with more right to speak and encourage students to self-evaluate, group self-evaluation, students' mutual evaluation, and group mutual evaluation. From these options, it can be seen that excellent English teachers do not focus on students' test scores, but on the development of students' various gains, experiences and comprehensive application abilities in the process of English learning. They evaluate students' English learning from a long-term perspective and meet the students' individual learning needs and future development. Excellent English teachers tend to take English tests as the main way of evaluating decision-making, which is also related to China's national conditions to a certain extent, and evaluation conferences and portfolios may not be suitable. Excellent English teachers let students selfevaluate and evaluate each other, students learn from each other, learn from each other, form healthy competition, and ultimately promote students' individual and group English learning. After the data analysis and the summarization of the text data, it is found that the teaching evaluation decisionmaking of excellent English teachers in colleges and universities mainly has the following characteristics: (1) Take the students' learning process as the center, and form a preliminary evaluation of students' learning. (2) Focus on reviewing the teaching plan and revise and improve the teaching plan. (3) To test the effect of the revised teaching plan as the core, test and evaluate the teaching effect again. From these characteristics, it can be seen that the teaching evaluation decisions of excellent English teachers in colleges and universities are always made around students. Moreover, teachers respond to the call of higher education reform, constantly reflect on and adjust teaching plans according to students' learning needs and teaching situations. Under the influence of certain educational values, teachers' knowledge, teachers' beliefs, and constantly formed practical knowledge, through the prediction, analysis, and reflection of teaching practice, make immediate evaluation decisions and adjust teaching. It tries to create a problem-solving and attractive teaching situation for students' learning as much as possible, flexibly adopts a variety of teaching methods and strategies, and strives for excellence. Through practice, the effect of the teaching plan is constantly revised and tested, and students are encouraged to participate more actively in classroom interaction, so that the teaching effectiveness is improved, and the teaching quality is also significantly improved. Because teachers' teaching decision-making is an implicit thinking process, the influence of internal factors on teachers is naturally greater than that of external factors. Given that English teachers' opinions and beliefs about the nature of language, roles of teachers and learners, and principles of teaching and learning, etc., subtly dominate teachers' decision-making. Therefore, teachers' belief is the primary factor affecting teachers' teaching evaluation decision-making. But only belief without corresponding knowledge is not enough, because teacher's knowledge is the basis for teachers to carry out teaching work and also the basis for teachers' teaching decision-making. Therefore, teachers' knowledge is a secondary factor affecting teachers' teaching evaluation decision-making. With beliefs and knowledge, teachers can make certain fixed or programmed decisions, but in the face of rapidly changing teaching situations, teaching cannot be carried out as planned, and students may raise unexpected questions or emergencies. Therefore, a certain amount of practical wisdom is also needed to help teachers adapt to changes, so that the classroom can be carried out smoothly. In a word, teachers' teaching evaluation decisions are influenced by teachers' beliefs, teachers' knowledge, practical wisdom, individual students, and teaching environment. The teacher's evaluation, modification, and adjustment of the teaching plan over and over again make the three stages of teaching plan decision-making, interactive decision-making, and evaluation decision-making inseparable, continuous, interlocking, and closely linked. Finally, teachers' teaching activities can be carried out smoothly, students can learn easily, and teachers can teach easily. This kind of evaluation and decisionmaking habit of continuous reflection, adjustment, and improvement of teaching is a treasure book for teachers to grow into excellent teachers. Therefore, teachers should pay attention to cultivating this good habit and cannot stay on the surface of reflection and fault reflection. Instead, the decision-making of teaching plan and teaching interaction should be sublimated through teaching evaluation and decision-making, so that these three can form a continuous whole, reciprocate and interact, and finally achieve the optimization of teaching effect. Conclusion In order to verify the necessity and feasibility of the evaluation system, this paper conducts teaching evaluation experiments in a targeted manner. Through teaching evaluation experimenters found that the evaluation value concept focusing on educational value is conducive to creating a democratic and equal evaluation atmosphere and is conducive to mobilizing the enthusiasm of the evaluation object for teaching or learning. By attracting stakeholders such as teachers, students, parents, and enterprises to participate in the evaluation of English teaching, it is conducive to coordinating the relationship between all parties and improving the reliability of the evaluation. The evaluation of English teaching is carried out on the evaluation objects including teaching content, teaching methods, and teaching resources, using a variety of evaluation methods such as formative evaluation, diagnostic evaluation, and summative evaluation. It is conducive to the objective and fair evaluation of the internal and external factors affecting the effect of English teaching. The use of manual + computer evaluation technology is conducive to improving the efficiency of English teaching evaluation. The combination of standardization and individualization of evaluation criteria is conducive to exerting the selection and identification function and the guiding and stimulating function of teaching evaluation. Using a combination of test-based and non-test-based assessment tools helps demonstrate the subject's knowledge and proficiency in the English language as well as non-English language proficiency. In a word, the establishment of the English teaching evaluation system should be based on the continuous response of the humanized evaluation concept of teachers and students' emotional appeal, which is a process of progress and continuous improvement. It absorbs various evaluation methods such as students and teachers, humanized evaluation, summative evaluation, qualitative evaluation, quantitative evaluation, and self-evaluation. According to the evaluation factors such as teaching attitude, improve the knowledge level of English and non-English language skills and the skill level of the evaluation object, so as to improve the reliability and validity of English teaching evaluation, and improve the quality of English learning. Due to some objective conditions and limited time, there are regrets and deficiencies: First, students are the main object of teachers' teaching evaluation decisions, whether we should still examine excellent English teachers and their teaching evaluation decisions from the perspective of students. Second, although the research has designed a questionnaire, the selected sample size is not large enough, so the research results and the scope of promotion are limited. Data Availability No data were used to support this study. Conflicts of Interest There are no potential competing interests in our paper. Authors' Contributions The author has seen the manuscript and approved to submit to your journal. 13 Wireless Communications and Mobile Computing	2022-03-21T15:09:51.274Z	2022-03-18T00:00:00.000	{ "year": 2022, "sha1": "702cbe06e85bcfda408811e9ab8b8608797f5dc6", "oa_license": "CCBY", "oa_url": "https://downloads.hindawi.com/journals/wcmc/2022/2999654.pdf", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "3e4fe3555bcb35da68e9417c20fc99c69f3bf643", "s2fieldsofstudy": [ "Education" ], "extfieldsofstudy": [] }
240038145	pes2o/s2orc	v3-fos-license	A prospective observational study on the predictability of Triple-D score versus Quadruple-D score in the success rate of extracorporeal shock wave lithotripsy of renal stones 1–2 cm in diameter Introduction: The aim of the study is to evaluate the clinical efficacy of Triple-D scoring system versus Quadruple-D scoring system for assessing stone-free rate (SFR) in individuals with renal stones measuring 1–2 cm in diameter after extracorporeal shock wave lithotripsy (ESWL). Materials and Methods: The study was conducted on 120 patients who presented to a tertiary care center in eastern India. Systemic random sampling technique was applied with a sampling interval of 2. Triple-D scoring system comprising of three computed tomography based metrics – stone dimension (volume), stone density (Hounsfield unit), and skin-to-stone distance (SSD) was done before ESWL. Stone location was included as an additional parameter to formulate Quadruple-D scoring system where an extra score was given for stones in the non-lower polar region. Stone-free status was assessed by plain abdominal radiography 3 weeks after ESWL. Results: In the study population, stone dimension, stone density, and stone location were positive predictors of SFR after ESWL whereas age, sex, and body mass index of the patients, laterality of the stone and SSD were not. The area under the curve of Triple-D and Quadruple-D scoring systems were 0.598 and 0.674. Conclusion: Triple-D scoring system has been successfully validated as the SFR showed a parallel increase with every positive component. The Quadruple-D scoring system with a simple addition of stone location can further facilitate the validation of Triple-D scoring by increasing SFR, keeping the calculation simple and easy to use. These findings support the incorporation of Quadruple-D scoring system over Triple-D scoring system. INTRODUCTION There are numerous factors which affect the stone-free rate (SFR) after extracorporeal shock wave lithotripsy (ESWL) such as the stone location and size, [1][2][3] composition, [4,5] Hounsfield unit (HU) as determined by computed tomography (CT), [6] intrarenal anatomy, [1,3] skin-to-stone distance (SSD), [7,8] and body mass index (BMI). [9] Triple-D scoring system was proposed by Tran et al. [10] Ichiyanagi O et al. proposed Quadruple-D scoring system as an extension of Triple-D scoring system. The aim of this study is to evaluate the clinical efficacy of Triple-D versus Quadruple-D scoring system for assessing SFR in individuals with renal stones measuring 1-2 cm in diameter after ESWL. MATERIALS AND METHODS In routine urological practice, renal and ureteral stones are very commonly encountered pathologies after urinary tract infection and disease of prostate. Endourological procedures such as percutaneous nephrolithotomy (PCNL) and ureteroscopy (URS) are now the preferred treatment modalities for renal stones. However, ESWL still remains a recommended treatment option for solitary renal stones of <2 cm in dimension. [1,11] A prospective observational study was conducted from April, 2019 to July, 2020 on 120 patients who presented to the outpatient department of a tertiary urology care center in eastern India. The sample size was calculated using the formula for descriptive study. One hundred and twenty participants were selected for the study using systemic random sampling technique with a sampling interval of 2 (two). The study was conducted as per guidelines laid down by the Declaration of Helsinki. The study protocol was approved by the institutional ethical committee. Inclusion criteria were (1) age more than 18 years, (2) urine culture-negative patients, (3) patient receiving ESWL for the first time for the targeted stone and (4) patient with no anatomical urinary tract abnormalities. The exclusion criteria were (1) distal urinary tract obstruction, (2) unavailable CT images before ESWL, (3) pregnant patients, (4) staghorn stones, (5) calyceal diverticular stones, (6) coagulopathy, (7) urinary tract infection, and (8) endourological/open procedures before ESWL. Before ESWL, plain abdominal radiography and helical non-contrast CT (NCCT) scan of the kidney, ureter, and bladder region (KUB) were done for evaluation of the renal stones. The coagulation profile and urine culture sensitivity were also done. Ellipsoid stone volume (SV) was measured using the formula SV = π/6 × (anteroposterior × transverse × craniocaudal diameters) in millimeters. [10,12,13] Stone density was measured in HU. SSD was calculated as the average distance from the body surface to a targeted stone at 0°, 45°, and 90° on NCCT. [7] Triple-D score was calculated as the sum of the number of components matching the cutoffs of <150 mm 3 for SV, <600 HU for stone density, and <12 cm for SSD as described by Tran et al. [10] Quadruple-D scoring system was defined as Triple-D score combined with the stone location (i.e., distribution). The location was allocated 0 point if the stone was placed at the lower calyx and 1 point if the stone was located at other sites. [14] The score would range from 0 (worst) to 3 (best) points and 0 (worst) to 4 (best) points in Triple-D and Quadruple-D scoring system, respectively [Tables 1 and 2]. Electromagnetic shockwave lithotripter, Dornier Compact Sigma, manufactured by Dornier MedTech systems GmbH was used. ESWL was performed with a gradual ramping up of shockwave energy at a fixed frequency rate of 60 shocks/min. The patients underwent just a single ESWL session as part of this study. Plain abdominal radiography of KUB was done 3 weeks after ESWL to assess the stone-free status [ Figure 1]. Statistics IBM SPSS statistics 26.0 developed by IBM, New York, USA was used to do all the statistical analysis. Student's ttest or Mann-Whitney U-test was used to compare continuous variables. Their correlations were assessed using Pearson's correlation analysis. Fisher's exact test and Chi-square test were used to analyze the cross charts between two categories. All P values were based on two-sided statistical analysis. P < 0.05 was considered statistically significant. RESULTS The study population of 120 patients was divided into two groups after fulfilling the inclusion and exclusion criteria. • Group A: Had stone-free status 3 weeks after ESWL • Group B: Had residual stone 3 weeks after ESWL. The age, sex, BMI, laterality of the stone, and SSD were not statistically significant for the prediction of stone-free status. Sixty-nine (57.5%) patients in the study population had stones at the pelviureteric junction, 14 (11.66%) patients had stones at the renal pelvis, 22 (18.33%) had stones in the lower calyx, 12 (10%) in middle calyx, and 3 (2.5%) in upper calyx, respectively. Fifty (65.78%) patients of Group A and 19 (43.18%) patients in Group B had stones at the pelviureteric junction, 10 (13.15%) patients in Group A and 4 (9.09%) patients in Group B had stones at the renal pelvis, 7 (9.20%) patients in Group A and 15 (34.09%) patients in Group B had stones in lower calyx, 7 (9.20%) patients in Group A and 5 (11.36%) patients in Group B had stones in middle calyx, and 2 (2.63%) patients in Group A and 1 (2.22%) patient in Group B had stones in upper calyx, respectively. Using the Fisher's exact probability test, P value is 0.014 (<0.05). Difference in the two groups with respect to stone location is statistically significant, making stone location an important predictor of success of ESWL. The mean ± standard deviation (SD) (in mm 3 ) ellipsoid SV is 396.44 ± 163.23 and 395.81 ± 227.52 in Groups A and B, respectively. Using the independent samples t-test, P value is 0.049 (<0.05). Hence, the differences in the SV in the two groups were statistically significant and an important predictor of success of ESWL. The mean ± SD (in HU) stone density was 724.28 ± 210.90 and 814.56 ± 190.63 in Groups A and B, respectively. Using the Student's t-test, P value is 0.001 (<0.05). The difference in the stone density between the groups was statistically significant and a positive predictor of ESWL success. The mean ± SD (Score) Triple-D score was 1.18 ± 0.58 and 0.93 ± 0.545 in Groups A and B, respectively. The P value is 0.026 (<0.05) using the Mann-Whitney U-test, so the difference in the Triple-D score in both the groups is statistically significant. The mean ± SD (score) Quadruple-D score was 2.09 ± 0.65 and 1.54 ± 0.79 in Groups A and B, respectively. Using the Mann-Whitney U-test, the P value is < 0.001 (<0.05), so the difference in the Quadruple-D score in both the groups is statistically significant. Triple-D and Quadruple-D scores can thus be used as important clinical assessment tools to predict the success rate of ESWL [ Table 3]. DISCUSSION In our study, we see that Triple-D score and lower pole location are independent predictors of SFR after ESWL for 1-2 cm renal stones. The SFR improved with increasing Triple-D and Quadruple-D score. The Quadruple-D score may be even more relevant than Triple-D score in making the clinical decision before choosing ESWL as a treatment modality. The receiver operating characteristic curve analysis revealed a low AUC of 0.598 in Triple-D score and 0.674 in Quadruple-D score for SFR prediction. In a similar study by Ichiniyag O et al., the AUC of Triple-D score was 0.596 and AUC of Quadruple-D score was 0.651. This may be because the SSD parameter, a component of both Triple-D and Quadruple-D scores, is not a statistically significant factor for discriminating stone-free or residual stones after ESWL. Contrary to our study, SSD and BMI, which are clinical indicators of obesity, have been reported as significant predictors of ESWL outcome in multivariate analysis. [15,16] SSD and BMI were not related to the SWL outcomes in the present study possibly because the study population comprised of mostly underprivileged and low socioeconomic status patients (mean BMI of 24.62 ± 1.06 kg/m 2 ), reflecting the racial background different from previous similar studies. [7,[15][16][17] Lower polar location of renal stones is a significant factor related to poor SFR after ESWL. An obtuse infundibular-pelvic angle, long lower calyx (<1 cm), and narrow infundibulum (<5 mm) are depicted as unfavorable factors for ESWL success. [1] However, these details were not incorporated in Quadruple-D score for ease of use in clinical practice. Increased stone burden, lower polar location, and increased SSD, all decrease success rate of ESWL and URS but has limited influence on PCNL outcomes. [3] Hence, for 1-2 cm renal calculi, stone and anatomical factors must be carefully examined when considering ESWL as the treatment modality. In a similar study, Ozgor et al. revealed the importance of stone location in addition to Triple-D score for predicting ESWL success in their multivariate analysis. [13] In our study, age is not a significant predictor of ESWL success rate. Contrary to this observation, age was reported as an independent predictor of ESWL outcome in multivariate analyses. [14,15,18] In another prospective study, [16] age and ESWL success rate reached a statistical significance in a univariate but not in multivariate analysis. Hence, age is not considered a parameter of Triple-D and Quadruple-D study. There are also other studies where age was not considered as having any significant impact on ESWL outcome. [15,[19][20][21] In a study correlating the age with ESWL efficacy, [22] it was seen that renal stones were difficult to fragment with ESWL in older patients than young patients. There is also a higher probability of renal hematoma after ESWL, which increased with age. Hence, age might have a negative impact on SFR. Many other nomograms exist for prediction of the successful outcome after ESWL, [2,10,15,21,23,24] but they are often too complex to calculate in clinical settings even though these have excellent outcomes. In a nomogram by Kim et al., manual scoring system was formulated using four to six variables on graphical chart in a CT-dependent or independent manner. Beside the four variables, sex, stone location, number, and maximal diameter, hydronephrosis grade and stone CT attenuation are included in the CT dependent nomogram. [24] This nomogram including Triple-D as well as Quadruple -D scoring system is practical and easy to use and remains externally validated. There are limitations in the present study. The infundibulo pelvis angle, infundibular length and width of lower calyx, and hydronephrosis were not assessed. Other limitations include relatively small number of patients. It is not clear how to extrapolate Triple-D and Quadruple-D score for ureteral stones. Further studies are needed to confirm the validity of the present findings. CONCLUSION In Indian patients with renal stones between 1 and 2 cm, Triple-D scoring system has been successfully validated as the SFR showed a parallel increase with every positive component of Triple-D scoring system. The Quadruple-D scoring system with a simple addition of stone location (non-lower polar vs. lower polar) can further facilitate the validation of Triple-D scoring by increasing SFR, keeping the calculation simple and easy to use.	2022-01-31T15:02:19.313Z	2021-10-26T00:00:00.000	{ "year": 2021, "sha1": "0602e52705f2c2342f404d169975650386fa9916", "oa_license": "CCBYNCSA", "oa_url": "https://doi.org/10.4103/ua.ua_1_21", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "fe616269ad4c7e715910a16f31b43739cb856307", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
43490654	pes2o/s2orc	v3-fos-license	Caspase-3 Is the Primary Activator of Apoptotic DNA Fragmentation via DNA Fragmentation Factor-45/Inhibitor of Caspase-activated DNase Inactivation* Caspase-3 initiates apoptotic DNA fragmentation by proteolytically inactivating DFF45 (DNA fragmentation factor-45)/ICAD (inhibitor of caspase-activated DNase), which releases active DFF40/CAD (caspase-activated DNase), the inhibitor's associated endonuclease. Here, we examined whether other apoptotic proteinases initiated DNA fragmentation via DFF45/ICAD inactivation. In a cell-free assay, caspases-3, -6, -7, -8, and granzyme B initiated benzoyloxycarbonyl-Asp-Glu-Val-Asp (DEVD) cleaving caspase activity, DFF45/ICAD inactivation, and DNA fragmentation, but calpain and cathepsin D failed to initiate these events. Strikingly, only the DEVD cleaving caspases, caspase-3 and caspase-7, inactivated DFF45/ICAD and promoted DNA fragmentation in an in vitro DFF40/CAD assay, suggesting that granzyme B, caspase-6, and caspase-8 promote DFF45/ICAD inactivation and DNA fragmentation indirectly by activating caspase-3 and/or caspase-7. In vitro, however, caspase-3 inactivated DFF45/ICAD and promoted DNA fragmentation more effectively than caspase-7 and endogenous levels of caspase-7 failed to inactivate DFF45/ICAD in caspase-3 null MCF7 cells and extracts. Together, these data suggest that caspase-3 is the primary inactivator of DFF45/ICAD and therefore the primary activator of apoptotic DNA fragmentation. Caspase proteinases drive apoptotic signaling and execution by cleaving critical cellular proteins solely after aspartate residues (reviewed in Ref. 1). Caspases exist as latent zymogens, but apoptotic death stimuli activate the initiator caspases, caspase-8 and caspase-9. Death receptors such as Fas induce caspase-8 activation via the adapter molecule Fas-associated death domain protein. Chemotherapeutic agents and UV irradiation cause release of mitochondrial cytochrome c, which then binds the adapter molecule apoptotic proteinase activating factor-1 (APAF-1), 1 and this complex along with adenine nucleotides promotes caspase-9 autoactivation. Once activated, initiator caspases in turn activate the executioner caspases, caspases-3, -6, and -7. The active executioners promote apoptosis by cleaving cellular substrates that induce the morphological and biochemical features of apoptosis (1). Besides caspase-3, caspases-6, -7, -8, -9, the cytotoxic T cell proteinase granzyme B, the calcium-dependent proteinase calpain, and the lysosomal proteinase cathepsin D may function during apoptosis (9 -12). These proteinases have all been suggested to participate in apoptotic DNA fragmentation; however, the mechanism(s) by which they promote DNA fragmentation remain unclear. Here, we examined whether these proteinases induced DNA fragmentation by inactivating DFF45/ICAD. We find that caspase-3 and caspase-7 are the only direct inactivators of DFF45/ICAD. However, endogenous levels of caspase-7 failed to inactivate the inhibitor or promote DNA fragmentation in vitro or in intact cells, suggesting that caspase-3 is the primary regulator of apoptotic DNA fragmentation via proteolysis of DFF45/ICAD. Drs. Guy Salvesen (The Burnham Institute) and Vishva Dixit (Genentech) provided plasmids encoding histidine-tagged caspases-3, -6, -7, and -8 (9,14). A construct encoding 6ϫ-histidine tagged caspase-9 was produced using the polymerase chain reaction and a topoisomerase-based cloning system (Invitrogen). Briefly, the coding region of caspase-9 was amplified from the pRSV-LacZ-caspase-9 plasmid (15) (Dr. Emad Alnemri, Thomas Jefferson University). The polymerase chain reaction product was subjected to agarose gel electrophoresis and purified by gel extraction (Qiagen). The gel-purified product was then cloned into the pTrcHis2TOPO vector according to the manufacturer's instructions. The fidelity of the construct sequence was confirmed by nucleotide sequencing. Cell-free Reactions-Cytosolic extracts from Jurkat cells and MCF7 cells were prepared as described (13,17). Cell-free reactions were conducted in the following buffers (as indicated): CAD buffer (10 mM HEPES-KOH, pH 7.2, 50 mM NaCl, 20% v/v glycerol, 2 mM MgCl 2 , 5 mM EGTA, 5 mM dithiothreitol, 1 mg/ml bovine serum albumin), calpain buffer (CAD buffer lacking EGTA and containing 2.5 mM calcium chloride and 0.1 mg/ml bovine serum albumin), and cathepsin D buffer (10 mM PIPES-KOH, pH 6.7, 50 mM NaCl, 20% (v/v) glycerol, 2 mM MgCl 2 , 5 mM dithiothreitol, 1 mg/ml bovine serum albumin). Since preliminary experiments demonstrated that calcium present in calpain buffer promoted DNA fragmentation even without calpain addition, we performed cell-free reactions in two steps. First, extracts were treated with proteinases or cytochrome c plus dATP as indicated in the figure legends. Aliquots of each reaction mixture were then withdrawn for caspase assay and Western blot analysis (13). Second, 10 6 rat liver nuclei and 5 mM EGTA were added to the extracts and the reactions then continued for the indicated times. DNA fragmentation was subsequently visualized via agarose gel electrophoresis and ethidium bromide staining (17). For the kinetic analysis described in the legend to Fig. 4, cell-free reactions were performed by mixing cytosolic extract, nuclei, and cytochrome c plus dATP in a single step as described in the figure legend. In Vitro Transcription and Translation-35 S-DFF45/ICAD was prepared from pBS-ICAD L using [ 35 S]methionine (Amersham Pharmacia Biotech) and a reticulocyte lysate transcription and translation system (Promega). Since functional DFF40/CAD can only be produced in the presence of DFF45/ICAD (3, 7), we included 200 ng of GST-DFF45/ ICAD in the transcription/translation reaction to prepare 35 S-pro-DFF40/CAD. Aliquots (2 l) of 35 S-DFF45/ICAD and 35 S-pro-DFF40/ CAD were incubated with proteinases in caspase buffer (20 mM PIPES, 100 mM NaCl, 10 mM dithiothreitol, 1 mM EDTA, 0.1% CHAPS, 10% sucrose, pH 7.2) as indicated in the figure legends and the products subjected to SDS-PAGE and autoradiography. In Vitro DFF40/CAD Assay-In vitro DFF40/CAD assays were performed as described (3). Briefly, pro-DFF40/CAD was prepared by in vitro transcription and translation in the presence of 200 ng of GST-DFF45/ICAD. Aliquots (2 l) of pro-DFF40/CAD were incubated with proteinases for 30 min at 30°C. 10 6 nuclei and EGTA (final concentration 5 mM) were added to the reaction mixture and after 2 h DNA fragmentation was analyzed via agarose gel electrophoresis and ethidium bromide staining (17). In control experiments, we determined that the reticulocyte lysate used for transcription and translation did not have DNase activity either in the absence or presence of proteinases or cytochrome c plus dATP. Caspases, Granzyme B, and Cytochrome C plus dATP Promote DFF45/ICAD Cleavage and DNA Fragmentation in a Cell-free Assay-We first assessed whether various apoptotic proteinases could promote DNA fragmentation in a cell-free assay. Proteinases were incubated with cytosolic extracts from Jurkat T cells and subsequently examined for DNase activity using rat liver nuclei as a substrate. We also examined whether the proteinase-treated extracts cleaved DFF45/ICAD and the fluorogenic caspase substrate DEVD-AFC. As shown in Fig. 1, all proteinases examined promoted DFF45/ICAD cleavage as demonstrated by Western blotting. However, not all cleavage events inactivated the inhibitor. Caspases-3, -6, and -7 produced DFF45/ICAD cleavage products of ϳ23 and ϳ12 kDa. Note that the ϳ23-kDa band likely represents intermediate cleavage products resulting from cleavage at one of the two caspase cleavage sites; whereas, the broad band at ϳ12 kDa probably represents a composite of the three DFF45/ICAD fragments observed when the inhibitor is cleaved at both caspase sites (2). The control lane shows a small amount of the intermediate cleavage products, likely due to background caspase activity. With granzyme B and cytochrome c plus dATP, only the ϳ12-kDa fragments were observed, suggesting cleavage had occurred at both caspase cleavage sites. By contrast, calpain prompted loss of DFF45/ICAD immunoreactivity and cathepsin D produced a ϳ19-kDa DFF45/ICAD fragment. Strikingly, only the caspases, granzyme B, and cytochrome c plus dATP elicited DEVD cleaving caspase activity (Fig. 1A), DFF45/ICAD inactivation, and DNA fragmentation (Fig. 1C). As in apoptotic cells (2), DNA fragmentation correlated with the production of ϳ12-kDa DFF45/ICAD fragments. Although calpain and cathepsin D induced DFF45/ICAD cleavage, they failed to inactivate the inhibitor, induce DNA fragmentation, or activate DEVD cleaving caspases. This suggests that DEVD cleaving caspase activity is not necessary for DFF45/ICAD cleavage, but is required for inactivation of the inhibitor and DNA fragmentation. We also examined whether the initiator caspases functioned in the cell-free assay. Caspase-8 (50 nM) induced DEVDase activity, DFF45/ICAD cleavage, and DNA fragmentation (not shown). Caspase-9 failed to promote these events; however, our caspase-9 preparation showed little activity against the fluorogenic caspase-9 substrate LEHD-AFC. This is probably due to lack of cofactors since caspase-9 demonstrates little activity in the absence of cytochrome c, dATP, and cyotsolic factors (APAF-1) (18). Caspase-3 and Caspase-7, but Not Other Apoptotic Proteinases, Promote DFF45/ICAD Inactivation and DNA Fragmentation in Vitro-We next asked whether the preceding apoptotic proteinases directly cleaved DFF45/ICAD or DFF40/CAD that was bound to the inhibitor. To accomplish this, we treated 35 S-DFF45/ICAD and 35 S-pro-DFF40/CAD with each proteinase and examined the products by SDS-PAGE and autoradiography. As shown in Fig. 2A, DFF45/ICAD was susceptible to each proteinase; however, the extent and pattern of DFF45/ ICAD cleavage varied with each proteinase. To determine whether DFF45/ICAD cleavage inactivated the inhibitor and released active DFF40/CAD, we treated pro-DFF40/CAD (prepared by in vitro transcription and translation in the presence of GST-DFF45/ICAD) with each proteinase and then examined DNase activity against rat liver nuclei. Strikingly, although each proteinase cleaved DFF45/ICAD, only caspases-3, and -7 inactivated the inhibitor, prompting release of active DFF40/CAD and DNA fragmentation (Fig. 2C). The initiator caspase, caspases-8, also failed to inactivate DFF45/ ICAD, even at concentrations up to 200 nM (not shown). DFF45/ ICAD inactivation and DNA fragmentation correlated with production of 24-, 22-, and 12-kDa DFF45/ICAD fragments that were only observed with caspase-3 and caspase-7 or when 35 S-DFF45/ICAD was incubated with cytochrome c plus dATPtreated cytosol (Fig. 2A). Although the other proteinases cleaved DFF45/ICAD, they failed to inactivate the inhibitor. Since granzyme B, caspase-6, and caspase-8 promoted DFF45/ ICAD inactivation and DNA fragmentation in the cell-free assay (Fig. 1), they probably function in this system via activation of caspases-3 and/or -7. Further Analysis of the Reaction of Caspase-3 and Caspase-7 with DFF45/ICAD and Pro-DFF40/CAD-We next examined the reaction of caspase-3 and caspase-7 with DFF45/ICAD in detail since these proteinases were the only direct inactivators of the inhibitor. As shown in Fig. 3A, when equal amounts (40 nM) of caspase-3 or caspase-7 were reacted with 35 S-DFF45/ICAD, caspase-3 cleaved the inhibitor more rapidly than caspase-7. After 30 min at 37°C, aliquots (40 g of protein) of each reaction mixture were analyzed for DEVD cleaving caspase activity (A) and subjected to Western blotting (12% gels) with anti-DFF45/ICAD antibodies (B). 10 6 rat liver nuclei and EGTA (final concentration 5 mM) were added to each reaction and the samples then incubated for an additional 2.5 h at 37°C. DNA fragmentation was then assessed by agarose gel electrophoresis and ethidium bromide staining (C). Unless indicated otherwise, reactions were conducted in CAD buffer. Note that DFF45/ICAD has a predicted molecular mass of 36.5 kDa based on amino acid sequence (2); however, the protein and its alternatively spliced short isoform run as bands of ϳ45 and ϳ38 kDa, respectively (2). Cleavage at the N-terminal caspase site gives predicted products of ϳ23.7 and 12.8 kDa; whereas, cleavage at the C-terminal caspase cleavage site yields products of ϳ24.8 and 11.7 kDa. Hydrolysis at both caspase sites therefore yields fragments of ϳ12.8, 12.0, and 11.7 kDa. In B, L and S indicate the long and short DFF45 isoforms, I represents the intermediate caspase cleavage products (ϳ23.7, 24.8 kDa), and the asterisk (*) denotes the ϳ12-kDa DFF45 fragments that correspond with DNA fragmentation. In A, error bars represent the S.E., n ϭ 3. In B and C, the data are representative of three independent experiments. 35 DFF45/ICAD cleavage was evident within 10 min following caspase-3 treatment and complete by 2 h. By contrast, caspase-7 did not produce detectable DFF45/ICAD cleavage until 2 h and only ϳ50% of the DFF45/ICAD was cleaved at 4 h. In concentration dependence experiments (Fig. 3B), caspase-3 also cleaved DFF45/ICAD more efficiently than caspase-7. Following a 30min incubation, DFF45/ICAD cleavage was detectable with as little as 8.8 nM caspase-3; however, DFF45/ICAD cleavage by caspase-7 was not detectable until the caspase concentration was 88 nM. FIG. 2. Limited proteolysis of DFF45/ICAD by caspase-3 and caspase-7, but not other apoptotic proteinases, releases active DFF40/CAD in vitro. To determine if DFF45/ICAD cleavage corresponded with release of active DFF40/CAD, we incubated various concentrations of caspase-3 or caspase-7 with pro-DFF40/CAD and rat liver nuclei and then monitored DNA fragmentation. As shown in Fig. 3C, caspase-3 induced detectable DNA fragmentation at 50 nM, whereas with caspase-7, DNA fragmentation required 150 nM caspase-7. Together, the data indicate that caspase-3 inactivates DFF45/ICAD and induces DNA fragmentation more efficiently than caspase-7. Caspase-7 Activation Does Not Induce DNA Fragmentation in Caspase-3 Null MCF7 Cells or Cytosolic Extracts-To determine whether endogenous levels of caspase-7 could cleave DFF45/ICAD and promote DNA fragmentation in the absence of caspase-3, we used caspase-3 null MCF7 cell extracts in a cell-free assay. The MCF7 cells lacked caspase-3 as demonstrated by Western blotting, although they contained caspases-6, -7, -8, -9, and APAF-1 at comparable levels to Jurkat cells (not shown). We treated the extracts with cytochrome c plus dATP and then examined DEVDase activity, procaspase-7 processing, DFF45/ICAD cleavage, and DNA fragmentation as a function of time. For comparison, we also analyzed these events in Jurkat cell extracts, which contain caspase-3. As shown in Fig. 4, cytochrome c plus dATP initiated procaspase-7 processing and the onset of DEVD cleaving caspase activity in Jurkat and MCF7 extracts. However, procaspase-7 processing occurred earlier and was more extensive in the Jurkat extracts. Procaspase-3 processing occurred with identical kinetics to procaspase-7 processing in the Jurkat extracts (not shown). The Jurkat extracts demonstrated ϳ10-fold greater DEVDase activity than the MCF7 cells at 2 h, likely due to the combined activity of caspase-3 and caspase-7. Al-though DEVDase activity declined with time in the Jurkat extracts, the Jurkat DEVDase activity remained substantially greater than the MCF7 DEVDase activity at all time points. Caspase-7 processing and DEVDase activity correlated with the extent of DFF45/ICAD cleavage and DNA fragmentation in both extracts (Fig. 4, B and C). In the Jurkat extracts, DFF45/ ICAD cleavage was complete by 2 h; whereas, even after 6 h, only ϳ50% of DFF45/ICAD had been cleaved in the MCF7 extracts. Strikingly, while DNA fragmentation had occurred by 2 h in the Jurkat extracts, little DNA fragmentation had occurred in the MCF7 extracts even after 6 h of incubation, despite caspase-7 activation and partial DFF45/ICAD cleavage. Thus, cytochrome c plus dATP initiate caspase-7 activation in the MCF7 extracts, but this does not inactivate DFF45/ICAD or promote significant DNA fragmentation. Similarly, we detected caspase-7-like DEVDase activity in extracts prepared from apoptotic MCF7 cells, but we detected no DNA fragmentation by agarose gel analysis or propidium iodide staining and fluorescence-activated cell sorter analysis (not shown). We next incubated caspase-3 with MCF7 extracts and nuclei to determine whether this proteinase could initiate DNA fragmentation. Fig. 4D shows that 125 nM caspase-3 induced DNA fragmentation in the extracts. Similarly, addition of exogenous caspase-7 to the extracts initiated DNA fragmentation, although this required a higher concentration than caspase-3 (not shown). Together, these data suggest that in the absence of caspase-3, endogenous levels of caspase-7 do not inactivate DFF45/ICAD. DISCUSSION In this paper, we examined whether various apoptotic proteinases could promote internucleosomal DNA fragmentation by inactivating DFF45/ICAD, thereby releasing active DFF40/ CAD. Of the eight proteinases examined, we find that only caspase-3 and caspase-7 are direct inactivators of DFF45/ ICAD. However, caspase-3 was the more efficient inactivator and endogenous levels of caspase-7 failed to release active DFF40/CAD, suggesting that caspase-3 is the primary activator of apoptotic DNA fragmentation. These findings confirm and extend those of Liu et al. (19) who recently demonstrated that caspases-3 and -7, but not caspases-6 and -8, inactivate recombinant DFF45, releasing active DFF40 (19). Additionally, FIG. 3. Caspase-3 inactivates DFF45/ICAD and activates DNA fragmentation more efficiently than caspase-7. In A, 35 S-DFF45/ICAD (2 l) was incubated with caspase-3 (40 nM) or caspase-7 (40 nM) for the indicated times at 37°C. Reactions were terminated by denaturing samples in Laemmli sample buffer and the products subjected to SDS-PAGE (12% slabs) and autoradiography. In B, 35 S-DFF45/ICAD (2 l) was incubated with the indicated concentrations of caspase-3 and caspase-7 for 30 min at 37°C. Samples were then denatured and analyzed by SDS-PAGE (12% slabs) and autoradiography. In C, pro-DFF40/CAD was incubated with 10 6 rat liver nuclei for 2.5 h at 30°C and the indicated concentrations of caspase-3 and caspase-7. DNA fragmentation was then assessed by agarose gel electrophoresis and ethidium bromide staining. The data are representative of three independent experiments. our work emphasizes the central importance of caspase proteinases, particularly caspase-3, as the principal mediators of apoptosis. Three lines of evidence support our finding that caspase-3 is the primary inactivator of DFF45/ICAD and thus the primary activator of apoptotic DNA fragmentation. First, only proteinases that initiated DEVDase activity in Jurkat extracts, which is due primarily to caspase-3 inactivated DFF45/ICAD and initiated DNA fragmentation (Fig. 1). Second, caspase-3 cleaved DFF45/ICAD and promoted DNA fragmentation more effectively than caspase-7, the only other direct DFF45/ICAD inactivator (Figs. 2 and 3). Third, caspase-3 null MCF7 cells failed to fragment DNA during apoptosis and cytochrome c plus dATP-activated MCF7 extracts did not inactivate DFF45/ICAD or promote DNA fragmentation, despite caspase-7 activation (Fig. 4). However, addition of caspase-3 to these extracts restored their caspase-dependent DNase activity. These findings are consistent with the delayed or absent apoptotic DNA fragmentation observed in caspase-3 null cells and mice (20 -23). Caspase-3 activation is therefore fundamentally important for DFF45/ICAD inactivation. Although caspase-7 initiated DNA fragmentation in a cellfree assay (Fig. 1) and inactivated DFF45/ICAD in an in vitro nuclease assay (Fig. 2), our data suggest that caspase-7 plays a secondary role in inactivating the inhibitor. Like caspase-3, caspase-7 prefers DEVD-based peptide substrates (1), however, caspase-3 cleaved DFF45/ICAD more effectively than caspase-7 (Fig. 3) in vitro and the low level of caspase-7 DEV-Dase activity generated in MCF7 cells and extracts (Fig. 4) did not inactivate DFF45/ICAD. This suggests that the two caspases cleave macromolecular substrates with different effi- FIG. 5. A model of apoptotic DNA fragmentation. The model emphasizes release of active DFF40/CAD via caspase-3 dependent inactivation of DFF45/ICAD and highlights how other apoptotic proteinases may intersect this pathway. Note that the initiator caspases and granzyme B function primarily via activation of executioner caspases. Active caspase-3 is the primary inactivator of DFF45/ICAD, but caspase-7, which may be localized to mitochondria, plays a secondary role in DFF45/ICAD inactivation. Caspase-3 in turn activates caspase-6, which aids in nuclear apoptosis via cleavage of nuclear lamins. Caspase-6 can also amplify caspase-3 activity via activation of procaspase-3. Cathepsin D and calpain apparently function independently of caspases and DFF40/CAD, although calpain could potentially initiate high molecular weight DNA fragmentation via release of mitochondrial apoptosis inducing factor. FIG. 4. Endogenous levels of caspase-7 cleave DFF45/ICAD inefficiently and fail to promote DNA fragmentation. Cytosolic extracts were prepared from Jurkat cells and caspase-3 null MCF7 cells as described in the text. Extracts (final concentration 12 mg of protein/ ml) were incubated at 37°C with buffer or cytochrome c (10 M) plus dATP (1 mM) and 10 6 rat liver nuclei. At the indicated times, the nuclei were pelleted by centrifugation and analyzed for DNA fragmentation (C). Aliquots of the supernatant (40 g protein) were analyzed for DEVD cleaving caspase activity (A) and subjected to Western blotting with antibodies against caspase-7, DFF45/ICAD, and actin (B). In D, MCF7 extracts were incubated with 10 6 rat liver nuclei and the indicated concentrations of caspase-3. After 2.5 h at 37°C, DNA fragmentation was assessed by agarose gel electrophoresis and ethidium bromide staining. In A, error bars represent the S.E., n ϭ 3. In B, C, and D, the data are representative of three independent experiments. ciencies and that endogenous levels of caspase-7 do not initiate DNA fragmentation. Furthermore, since active caspase-7 localizes to mitochondrial and microsomal membranes (24), the caspase may be physically sequestered from DFF45/ICAD and therefore unable to react with the inhibitor. Overall, the data suggest that caspase-7 is not a physiologic inactivator of DFF45/ICAD and that critical caspase-7 substrates remain unidentified. Caspase-6 and granzyme B induced DFF45/ICAD inactivation and DNA fragmentation in Jurkat extracts (Fig. 1); however, although they cleaved DFF45/ICAD, they failed to inactivate the inhibitor in an in vitro nuclease assay (Fig. 2). Since these proteinases effectively activate caspase-3 (16,25), these results suggest that they inactivate DFF45/ICAD via caspase-3 activation. Non-activating DFF45/ICAD cleavage was also observed with calpain and cathepsin D (Figs. 1 and 2); however, since these proteinases do not activate caspases ( Fig. 1 and Ref. 13), they were unable to initiate DNA fragmentation in a cellfree assay (Fig. 1). Caspase-6, granzyme B, calpain, and cathepsin D probably cleave DFF45/ICAD at or near the C-terminal caspase cleavage site (Asp 224 ) since DFF45/ICAD mutants that can only be cleaved at this site retain DFF40/ CAD inhibitory activity (6). Non-activating cleavage of DFF45/ ICAD by these proteinases could be synergistic with activating cleavage events mediated by caspase-3 or caspase-7 if the partially cleaved inhibitor is more susceptible to cleavage by caspases-3 and -7. Alternatively, non-activating cleavage might alter interaction of DFF40/CAD with cofactors such as histone H1 (19) or target the endonuclease for destruction. These possibilities are currently under investigation. In summary, our data indicate that caspase-3 is the primary inactivator of DFF45/ICAD and suggest that proteolytic pathways that induce apoptotic internucleosomal DNA fragmentation must involve this proteinase. Fig. 5 summarizes our findings and presents a model of apoptotic DNA fragmentation. In this model, the initiator caspases, caspases-8 and -9, promote apoptotic signaling and activate the executioner caspases, which in turn degrade apoptotic substrates. Similarly, granzyme B functions primarily to activate executioner caspases and not to cleave apoptotic substrates. Studies demonstrating that granzyme B activates caspase-3 more efficiently than caspase-8 (16) and the lack of DNA fragmentation observed in the absence of caspase activation during granzyme B-mediated apoptosis (9) support this hypothesis. Once activated, caspase-3 plays a central role in apoptotic DNA fragmentation by inactivating DFF45/ICAD, thereby releasing active DFF40/ CAD. Caspase-3 also activates caspase-6 (26), which in turn promotes nuclear apoptosis by degrading lamins (27). Caspase-6 can also activate procaspase-3 (25), providing the opportunity for amplification of caspase-3 activity. Caspase-7, which may be localized to mitochondria and microsomes plays a secondary role in DNA fragmentation as a back up DFF45/ ICAD inactivator. By contrast, calpain and cathepsin D do not activate caspases and function downstream or independently of caspase cascades. However, calpain, which localizes to mitochondria during apoptosis (28), could potentially induce high molecular weight DNA fragmentation via release of mitochondrial apoptosis inducing factor. Although lysosomes release cathepsin D during some forms of apoptosis (29), this proteinase does not inactivate DFF45/ICAD and its cellular substrates remain undefined. Thus, many proteinases may function in nuclear apoptosis, but caspase-3 is the key determinant of DFF45/ICAD inactivation and apoptotic internucleosomal DNA fragmentation.	2018-04-03T05:36:14.769Z	1999-10-22T00:00:00.000	{ "year": 1999, "sha1": "fa72ac40f313aace1fc3f83853da2ec51e9fac45", "oa_license": "CCBY", "oa_url": "http://www.jbc.org/content/274/43/30651.full.pdf", "oa_status": "HYBRID", "pdf_src": "Highwire", "pdf_hash": "b9945b418c8fd620cb071dd922d2ce4a4aaaeb24", "s2fieldsofstudy": [ "Biology", "Chemistry" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
244250352	pes2o/s2orc	v3-fos-license	Evaluation of Infiltration Rainwater Drainage (IRD) System with Fully 3-D Numerical Simulation Approach : Water scarcity can mean scarcity in availability due to physical shortage, or scarcity in access due to the failure of institutions to ensure a continuously regular supply or due to a lack of adequate infrastructure. Water scarcity will be exacerbated as rapidly growing urban areas place heavy pressure on water resources. To solve these problems, various solutions have been applied, but a fundamental solution has not been applied. Recently, a researched and developed infiltration rainwater drainage (IRD) system is being applied with consideration of its applicability. In this study, features of surface runoff and infiltration according to various flow patterns were analyzed using a three-dimensional CFD (Computational Fluid Dynamics) model for calculating water flow in the IRD system. To estimate the optimal setup, a permeability test and scaled model simulation were performed. The runoff characteristics of the IRD system with respect to rainfall intensity and duration were analyzed with dimensionless variables. With the prototype model, the drainage characteristics of the IRD system were analyzed over time using the hydrological curves. From the simulated results, it was found that the IRD system analyzed in this study was appropriate in the field by comparative analysis with the existing system based on peak runoff, internal storage, and lag time. Therefore, by applying the IRD system in the future, it is expected that the IRD has benefits, such as delayed lag time, surface runoff decrease, and an attenuation of the peak runoff. Introduction Uncertainty in rainfall patterns due to climate change, urbanization, and industrialization, water security in the urban area has been continuously and globally considered as an important issue. In modern urban pavement, bituminous materials such as asphalt and impervious materials, such as concrete, are used, and a continuous increase in the runoff coefficient was expected. Numerous studies have analyzed the increase of surface runoff and various problems resulting from it with statistical analysis results using water level, rainfall distribution, and population density known to increase the risk [1][2][3]. In particular, changes in urban runoff characteristics cause a decrease in the efficiency of existing rainwater drainage systems, which are known to be a representative cause of various urban water circulation problems due to a decrease in the groundwater level. Global warming due to the climate change can bring the anomalies in the precipitation in the world [4]. In particular, the annual change in rainfall in terms of rainfall intensity in South Korea from 1961 to 2003 was studied by Kim et al., (2005). It was found that the rainfall intensity exceeding 20 mm/h gradually increases and the possibility of developing into a heavy rain further increases [5]. In addition, if the climate change and the increase of rainfall intensity continue, the total rainfall on the Korean Peninsula will continue to increase by the end of the 21st century and it was predicted to increase by about 25% [6]. The impervious land area in Korea has continuously increased over the past four decades, and it is now about 8.0%. Further, in Seoul, the biggest city in Korea, more than 54% of the area was impervious, so rainwater drainage efficiency was very low (Figure 1). Figure 1. Increase of Impermeable Area in South Korea [7]. Based on the statistical data for the last 7 years, it was also found that the impermeable area ratio of major cities has increased, except for some island areas (Figures 2 and 3). Therefore, Park and Shin (2014) conducted a research that the maximum outflow reduction effect of 15% of the total outflow volume and 18% of the peak flow could be expected by controlling the impervious pavement of application site [8]. A rainwater drainage system must have the ability to respond to hydraulic, hydrological, and environmental issues, and problems that are compounded by continuous urban development and climate change. The currently installed conventional rainwater drainage system is a method of draining the surface runoff flowing along the ground surface through grating-type rainwater harvesting on the outside of sidewalks and roads. However, its efficiency was rapidly reduced, due to the illegal dumping of garbage, the inflow of floating debris, and the installation of illegal covers in commercial facilities, which is a structural disadvantage ( Figure 4). In order to overcome the disadvantages of the conventional rainwater drainage system, an Infiltration Rainwater Drainage (IRD) system was developed, and various studies of evaluation and improvement were conducted. The IRD system was one which installed pavers with optimized permeability to utilize the porous volume and underground storage space of pavement materials and reduce surface runoff. In general, this system basically consists of a surface layer, a sub-base layer, and a filter layer [9]. The IRD system is known to have the effect of reducing peak discharge and flood level, and also has ecofriendly functions, such as securing groundwater refilling and mitigating the urban heat island phenomenon [10]. Despite these advantages, the IRD system was not yet widely applied. In particular, it was found that permeable pavement had a physical limitation of the drainage efficiency due to contaminants and pore constriction [11]. Because permeable pavement has a relatively low strength limit compared to pavement materials, such as concrete and asphalt, it was limitedly applied to parking lots, sidewalks, and bicycle paths where the load was light [12]. Therefore, this study was conducted to quantitatively evaluate the drainage performance of the IRD system, which is being actively researched and developed recently to solve the water circulation problem in urban areas caused by climate change. Among the existing numerical studies for the evaluation of the IRD system, the studies that applied the one-dimensional numerical analysis method usually proved the effect of reducing peak discharge due to IRD infiltration through flood level analysis in river basins. On the other hand, existing studies using the three-dimensional, numerical simulations have a limitation of calculating the surface runoff and infiltration separately and then merging the results. Consequently, previous simulations cannot calculate the runoff over time accurately. Therefore, we conducted fully 3-D numerical simulations of the permeability test model, a scaled-down model, and a full-scale prototype model of the IRD system, and evaluated its performance using a hydrologic curve. First, the permeability to determine the storage capacity and retention function were calculated by modeling two cases of permeability experiments for the design of the drainage layer of the IRD system. Second, the applicability of the IRD system with respect to various rainfall intensities and rainfall duration was verified. With the prototype model, the drainage performance of the system per unit catchment area according to the design factors (variables) was quantitatively compared and analyzed. Finally, a simulation was conducted using an integrated model to understand the drainage performance of the IRD system for a unit catchment area. In addition, the surface runoff and infiltration discharge were verified during the rainfall duration. The performance of the IRD system was quantitatively analyzed by comparing the simulation results and runoff characteristics of the existing rainwater drainage system. Im et al. (2007) conducted permeability tests using five types of pavement materials such as concrete pavement, asphalt pavement, soil, grassland, and permeable pavement. It has proven its suitability as a packaging material for rainwater runoff reduction facilities by analyzing that it secures a penetration rate of about 93% or more until it occurs below the rainfall intensity [13]. Im et al. (2009) noted that it was difficult to quantitatively represent the capacity of the infiltration-storage system because it showed a large difference depending on the engineering specifications of the soil and the Antecedent Precipitation Index (API). It was analyzed that the infiltration rate and storage were constant when the rainfall intensity was over 200 mm/h by analyzing the accumulated infiltration and storage ratio. In addition, the permeability of the permeable paver and the conventional paver were compared to prove that the permeability was kept constant [14,15] by analyzing the permeability test results. Köhne et al. (2011) performed simulations by combining the two-dimensional numerical model Hydrus-2D and AMAL-GAM to simulate surface runoff and infiltration [16]. Belheine et al. studied, using numerical tri-axial tests, the micro-mechanical properties of the numerical material, which were calibrated in order to match the macroscopic response of the real material. Based on the local behavior law, numerical simulations were carried out under the same conditions as the physical experiments (porosity, boundary conditions, and loading). They expected the fully 3-D numerical simulation can be helpful for further rainfall drainage management [17]. The IRD system was more efficient than the existing runoff reduction facilities because of its increasing infiltration rate. Such a result was a guaranteed application of IRD as part of runoff-reducing facilities. Therefore, the IRD system was expected to be viable for practical application with subsequent research for improvements [18][19][20][21][22][23][24]. Park et al. analyzed the efficiency and flow pattern of the permeable concrete and block pavement and suggested the ratios of permeable concrete and block pavement with experimental analysis [25]. These previous research results were classified as a completely integrated model. As a method of simulating and merging surface runoff and infiltration simulation results, it has a limitation that it cannot present runoff characteristics over time, so integrated modeling was required. Governing Equations For numerical simulation, it was necessary to convert the size, position, and shape of an object into grid materials. In general, the most widely used grid generation techniques are the Finite Element Method (FEM) and the Finite Volume Method (FVM). In this study, surface runoff, infiltration, and subsurface flow were simulated simultaneously by applying the FVM method. The numerical modeling was conducted using the Reynolds Averaged Navier-Stokes model (RANS) to simulate the surface runoff and its turbulent components. The porosity of the permeable pavement varied 20-40%, which was very large compared to the commonly applied packaging materials. It was difficult to design and perform simulations assuming that the flow inside the inner porous material is laminar due to the characteristics of the permeable pavement. Therefore, this study used a directional loss model developed based on Darcy's law to simulate the internal flow of a porous material. To simulate surface runoff and infiltration over time, flow analysis of each flow region was performed simultaneously with two analysis targets, and the runoff characteristics over time were simulated. First, a permeability test was simulated for the design of the drainage layer. The optimal permeability, which was applicable to the IRD system, was calculated. In order to represent the drainage performance of the unit drainage system with the real-scale model, the permeability of the pavement and the runoff with respect to the surface slope were classified with respect to the time, and the peak flow rate and the maximum runoff volume were compared. In general, the turbulence model can be expressed as the sum of the mean and fluctuation components of the turbulence. The unsteady Navier-Stokes equation was modified to obtain the following RANS model from the continuity and momentum equations [26]. where U is flow velocity of fluid and t is time step. is density of fluid and is momentum source. is the shear stress. Hence, the quantity of ∇ • ⨂ can be represented by . In this study, k-model was applied and the equation is as follows: where k is kinetic energy and is kinetic energy dissipation rate. is turbulence viscos- . is the turbulence model constant for the k equation. where and are the Reynolds stress model constants, and is turbulence production due to viscous and buoyancy forces. In the case of the full buoyancy model, the definition of is as follows: where is the gravitational constant (=9.81 m/s 2 ). is constant. Flows in Permeable Materials Applying a loss model developed based on Darcy's law, it was possible to calculate efficiently by preventing unnecessary grid generation. The form of Darcy's model applied to this simulation is as follows: where is dynamic viscosity, is permeability, and is the empirical loss coefficient (laminar flow: / and turbulent flow: /2). Model Setup of Geometry Data CFD simulations can be divided in three main steps, (a) pre-processing (definition of the region of interest, numerical mesh generation, initial and boundary conditions definitions, mathematical modeling and numerical schemes definitions); (b) simulation solving (numerical solution of the transport equations system); and (c) post-processing (treatment and analysis of results). Procedure of the CFD modeling using Ansys CFX is depicted in Figure 5. We created geometry by importing an existing DesignModeler in ANSYS package. Grid size of three-dimensional mesh was setup with 5 mm. Error Analysis The method of calculating the mass error (Merror) of the model is as follows. where Vpre is volume of precipitation and Vsur is volume of surface runoff. Permeability Test Results from Previous Study It was possible to calculate the permeability of the permeable covering material with a permeability test. From the similar test results, Figures 6 and 7, we applied two permeability values (Kperm= 1.0 × 10 −9 m 2 and 1.0 × 10 −10 m 2 ) [28]. The model of the test was combined with water body and soil part ( Figure 4). Boundary conditions of permeability test were summarized in Table 1. Permeability Test The validation of the model was performed by analyzing mass conservation. The discharge calculated in the upper part of the model and the infiltration mass in the lower part were compared for the two test conditions as shown in Figure 8 and Table 2. As a result of the simulations of the test, the peak infiltration was calculated to be 9.5 × 10 −3 and 8.8 × 10 −4 m 3 /s and each peak occurs at 0.067 and 0.78 s, respectively. The maximum bottom displacement was simulated as 6.6 × 10 −5 m 3 /s and 6.7 × 10 −5 m 3 /s at 7.25 and 71.3 s, respectively. The total drainage performance was simulated as 2.3 × 10 −2 and 2.4 × 10 −2 m 3 for the bottom drainage, respectively. In the case of Pe-10, about 3.53 % of increase was observed. However, the total drainage time was 200 s and 1000 s, almost five times increased. With the relationship between the amount of drainage and the total drainage time, the permeability of Pe-10 (=1.0 × 10 −10 ), which had relatively excellent retardation effect and internal storage performance, was determined to be an appropriate permeability to the drainage layer. Model Setup of Scaled Model Simulation The scaled model was a combination of a permeable layer and a drainage layer. It was designed and performed to identify surface runoff characteristics of the IRD system with respect to rainfall intensity and duration. There were five rainfall intensity cases, 10~200 mm/h, and three rainfall durations, 100, 200, and 600 s, respectively. The simulation conditions are summarized in Table 3 with 0.1 s of computational time step, ∆ = 0.1 s. -I1D2 200 Sc-I1D6 600 Sc-I3D1 30 100 5 Sc-I3D2 200 Sc-I3D6 600 Sc-I5D1 50 100 3 Sc-I5D2 200 Sc-I5D6 600 Sc-I10D1 100 100 2 Sc-I10D2 200 Sc-I10D6 600 Sc-I20D1 200 100 1 Sc-I20D2 200 Sc-I20D6 600 As for the specifications of the permeable layer, the permeability coefficient of an optimized value as a result of a previous research, conducted based on the samples collected from a field observation [29], was applied. Three layers were setup for the numerical simulation (Figure 9). Table 4 shows the total amount and maximum values of surface runoff and infiltration. The cases of the mass error of 0.5% or less were used for the analysis. The rainwater drainage through surface runoff occurs predominantly for all cases. Figures 10-12 show the dimensionless runoff for the same rainfall time by dividing the infiltration and surface runoff by the total runoff on the y-axis, and dividing time by the total rainfall duration on the x-axis. They also show the drainage characteristics in the case of the same duration with dimensionless variables. In the case of the scaled model, 90% of the total runoff is surface runoff, because the volume of porous media is not enough to hold infiltrated water under surface. Figures 10-12 show that the dimensionless surface runoff remains constant over time and the dimensionless infiltration increases. Therefore, it was revealed that there is an effect of reducing surface runoff if sufficient porous media to hold infiltrated water. When the product of the total rainfall and the duration is the same, that is, when the rainfall depth (I × t, mm) from Table 3 (2.78 and 5.56 mm) is the same, the runoff characteristics according to the rainfall intensity of the IRD system were analyzed as shown in Figures 13 and 14. They represent infiltration and surface runoff with dimensionless variables, respectively. In the case of a rainfall depth of 2.78 mm, the infiltration scale increases by up to 16.8% and an average of 10.8% as the rainfall intensity increase. In case of rainfall depth of 5.56 mm, the surface runoff scale decreases by up to 23.7% and 43.4%. This reversal of the infiltration scale was analyzed to be caused by insufficient volume of porous media compared to rainfall. Therefore, installing the IRD system, it is necessary to design appropriately considering the amount of rainfall and the duration of rainfall. Model Setup of Prototype Simulation Analysis of the prototype model was performed to quantitatively represent the drainage performance of the IRD system according to the design catchment area. In this model, block drainage discharge, drainage layer discharge, drainage hole, surface runoff, and infiltration were analyzed with respect to time. For the rainfall intensity applied in this simulation, 210 mm/h, the 200-year frequency of rainfall intensity in the central region of the Korean Peninsula, was applied. The design catchment area of the model was the same as the specifications of the IRD system actually built in Incheon city, South Korea. Figure 15 shows the schematic design of catchment area. For the validation of the model, it was designed based on the domestic constructing regulations [29]. According to MOLIT, in the case of a cross slope 2% is the construction standard, and a maximum of 5% is allowed. In order to simulate the drainage performance according to the transverse slope and permeability of the drainage layer and pavement, simulations were performed by applying permeable blocks of different permeability and compared to the results of application of existing forms. We selected the part from the catchment area in the field as the application model ( Figure 16). With four values of permeability and three cases of slope values including no slope (0%, 3%, and 5%), conditions of numerical simulation were setup as summarized in Table 5. We built the four kinds of drainage system model for the numerical simulation including normal type ( Figure 16). To improve the accuracy of the simulation of the prototype model, the mass conservation error of each case for the time interval and grid size was analyzed (Table 6). Based on the result that the error was less than 8%, the time interval and grid size were determined as shown in Table 6. Table 7 shows detailed results of runoff characteristics of the IRD system according to the various structures. The time of occurrence of the maximum infiltration was simulated to be 176 s and 112 s later, respectively. The peak of surface runoff exists at 587 and 600 s. When Pr-Pe4S10 designed as a multiple-structure was applied, the retention effect was demonstrated by slowing the time of maximum surface runoff through relatively fast maximum penetration compared to the cases of control group. It was found that the maximum infiltration flow rate was reduced by 81.21% compared to Pr-Pe0S10, an impervious drainage system. It can be concluded due to these changes in runoff characteristics, that when the IRD system with a multiple structure was applied to cities and watersheds, surface runoff decreases with retention effects by infiltration ( Figure 17). The hydrograph of the various structure cases showed that the runoff characteristics of the initial rainfall duration of 600 s were significantly changed according to the surface slope. Table 8 shows the runoff characteristics for each case according to the cross-slope. The maximum penetration occurred at 2.1 × 10 −3 m 3 /s, 112 s in the case Pr-Pe4S10 with 0% cross-slope. In addition, the maximum surface runoff reduction rate according to the slope was 69~81%, and the deviation according to the cross slope was about 12%. The cross slope was analyzed as a major design factor of the IRD system. The maximum surface runoff was simulated as 3.93 × 10 −4 m 3 /s, 7.29 × 10 −4 m 3 /s, and 7.99 × 10 −4 m 3 /s, respectively, as the cross slope increased, indicating a proportional relationship between the slope and the maximum surface runoff. In case of total infiltration, 0.923 m 3 , 0.742 m 3 , and 0.756 m 3 were simulated. Thus, the design of an excessive cross slope may cause an increase in the burden on the drainage system during high-intensity heavy rain ( Table 8). The surface runoff and infiltration of the IRD system with respect to the permeability of the pavement were respectively shown in Figure 15. In the case of the IRD system, it was found that the surface runoff amount was reduced significantly (Figure 18). Table 9 shows the maximum surface runoff, infiltration, and the timing of each occurrence. The maximum permeability occurs in Pr-Pe2S10, which had the largest permeability, but the decrease in the maximum surface runoff occurred in Pr-Pe4S10, which had the least permeability. In the case of Pr-Pe3Sl0, the time at which the maximum surface runoff occurs increases by 20 s, so it can be concluded that the urban water circulation problem can be solved if the delay time was designed with the permeable design. Figure 16 shows the surface runoff and infiltration of permeable rainwater drainage systems with respect to the permeability of the pavement. When the IRD system was applied, it was simulated that the surface runoff was reduced significantly (Figure 19). Based on the maximum surface runoff and rainfall duration, the runoff of each model was shown in dimensionless form ( Figure 20). As a result of the dimensionless comparison, it was possible to prevent the performance degradation of drainage facilities due to rapid surface runoff by applying the IRD system. In addition, it was found that the penetration pattern varies greatly depending on the design factor. The infiltration rate and surface runoff rate of each drainage system for total rainfall were calculated as shown in Table 10. Up to 71% of the precipitation applied to the design catchment area can be drained through infiltration with the IRD system. The detailed runoff, internal storage, and retention rates of each drainage system are compared as summarized in Table 11. In the case of Pr-PE1S10 designed with a simple drainage layer structure, the retention rate was simulated as 94.23%. The single drainage layer design was a relatively unfavorable condition for rainwater drainage compared to the complex structure design. As a result of simulation according to the permeability of Time/Rainfall duration the pavement when designing multiple structures, the highest total amount of infiltration was calculated as 1.11 m 3 for Pr-Pe2Sl0 due to the high permeability of the pavement. As a result of simulation of Pr-Pe4Sl0, Pr-Pe4Sl3, and Pr-Pe4Sl5, the infiltration amount decreased with the increase of the transverse slope. As shown in Table 9, this was identified as a phenomenon caused by an increase in the surface runoff with an increase of the crossslope. As a result of the overall simulation, the displacement was plotted with respect to time. As a result of the simulation, it was found that the outflow characteristics changes according to the design factors of the IRD system. The maximum inflow rate reduction effect of more than 50% was found when the IRD system was applied ( Figure 21). The maximum peak flow reduction rate was compared to the existing drainage system. The deviation of peak flow according to the design factor was 13% according to the structure, 10% according to the permeability, and 12% according to the slope. In other words, the deviation due to the structure of the drainage system was the most significant. The IRD system with a single structure drainage layer stores 99.4% of the total rainfall inside due to the permeability of the drainage layer. The reduction rate of peak discharge was shown in Figure 22. Compared to the case of Pr-PE0Sl0 (impervious pavement case), all the cases of the IRD reduce peak flow. The desirable reduction of the peak can be achieved combining cross section slope and permeability. Pr-Pe1Sl0 Pr-Pe2Sl0 Pr-Pe3Sl0 Pr-Pe4Sl0 Pr-Pe4Sl3 Pr-Pe4Sl5 Figure 22. Comparison of IRD system prototype model. Discussion From the result of the simulation of the previous permeability test, the drainage layer with a permeability of 10 −10 m 2 was chosen as the optimal permeability value. The scaled model was built for the IRD system test and its applicability was verified by applying various rainfall intensities and rainfall durations. As the rainfall duration increased, the surface runoff remained relatively constant with an average of 0.941, 0.942, and 0.943. Under the same rainfall depth condition, the infiltration was reversed when the volume of porous media was not sufficient. In the results of the scaled model simulation, the dimensionless values of surface runoff were almost constant at about 0.94 despite the change in rainfall duration. In the case where the product of total rainfall and duration is the same, that is, where the rainfall depth is the same, the runoff characteristics according to the rainfall intensity of the IRD system were identified. That is, in the case of a rainfall depth of 27.6 mm, the penetration result increased by 16.8% on average and 10.8% on average as rainfall intensity increased. For a rainfall depth of 55.5 mm, the magnitude of surface runoff decreased by up to 23.7% and up to 43.4%. As such, it is judged that the result that the infiltration result was calculated in the opposite way was because the infiltration area was not sufficiently secured compared to the amount of rainfall. Therefore, when constructing the IRD system, it is necessary to design the application site considering the rainfall amount and rainfall duration. Prototype modeling cases (Pr-Pe2Sl0, Pr-Pe3Sl0, and Pr-Pe4Sl0) were simulated with respect to the permeability of the permeable block, respectively. As a result of the simulation, the total storage rates were simulated to be 88.28, 31.37, and 94.78%, respectively. The permeability of the permeable block of Pr-Pe3Sl0 was the second biggest one among three of permeability variation cases. Although Pr-Pe3Sl0 had the second biggest permeability, it had the least retention rate among the three cases. Through the analysis of the relationship of permeability-total drainage time, in the case of Pr-Pe3S10, rainwater stagnated inside the permeable block and in the drainage layer was effectively drained after rainfall. This design can maximize the retention effect of the infiltration type rainwater drainage system that collects and drains rainwater slowly by using the internal pore internal storage function during rainfall. If the IRD system was designed with the test Peak outflow reduction ratio condition of Pr-Pe3Sl0 (permeable block with no cross slope), infiltration was equivalent to 62% of the total rainfall and 69% was continuously drained up to about 1,200 s under a heavy rain condition. It was concluded that the IRD system can be reduce the burden on the municipal drainage and the quantitative evaluation with 3-D integrated numerical modeling method suggested in this study. Conclusions Previous numerical studies have analyzed behavior of surface runoff and infiltration in the permeable pavements, but few studies used only 3-D numerical simulation or coupled modeling after separately analysis. This study was conducted to quantitatively identify the drainage performance of the IRD system with 3-D numerical simulation. Variables such as peak flow rate, peak flow rate reduction rate, internal storage amount, and lag time were analyzed. Compared to the existing system, the relative advantage and applicability were evaluated with a 3-D CFD modeling. The performance of the IRD system could be evaluated with the procedure suggested in this study. The 3-D CFD modeling, which includes surface runoff and internal infiltration, is expected to be applicable to the optimum design with respect to the dimension and the characteristics of the drainage area and rainfall pattern. The construction materials, such as blocks and sand, used for the drainage system, will vary with respect to construction conditions. While physical modeling will provide accurate evaluation for the various conditions, it takes time and effort to perform the physical modeling. The CFD modeling calibrated with physical model will further improve the accuracy and efficiency of the evaluation the IRD system under the various conditions. With HPC resources, this approach and procedure are very useful for the evaluation of rainfall drainage in urban area.	2021-10-18T17:40:43.289Z	2021-10-01T00:00:00.000	{ "year": 2021, "sha1": "bb886170d99176025771dcb248be5480cf90744b", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2076-3417/11/19/9144/pdf", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "8eb88c1baca0959166d6a415abcc8d0fdce32bc2", "s2fieldsofstudy": [ "Engineering" ], "extfieldsofstudy": [ "Environmental Science" ] }
250334592	pes2o/s2orc	v3-fos-license	Overview of thermodynamic properties for Reissner-Nordstrom-de Sitter spacetime in induced phase space Since the black hole and cosmological horizons in the de Sitter (dS) spacetime with the Reissner-Nordstrom (RN) black hole are not independent to each other, which is caused by the gravitational effect, the interplay between two horizons should be considered. Based on this, by introducing the interactive entropy the RN-dS spacetime is analogous to a thermodynamic system with various thermodynamic quantities, in which the thermodynamic laws still hold on. In our work, the thermodynamic properties of the RN-dS spacetime are mapped out in the induced phase space, which are similar to that in AdS black holes. The phase transition of the RN-dS spacetime between the high-potential and the low-potential black hole phases is observed. Compared with an ordinary thermodynamic system, the similar behaviors about the Joule-Thomson expansion and the critical exponents are also checked out. Finally, the scalar curvature of two existent phases are presented to reveal the underlying microstructure and nature of phase transition in the RN-dS spacetime, which opens a new window to investigate the dS spacetime with black holes from a observational perspective. I. INTRODUCTION With the development of theory, people found there is a fundamental relationship in gravity, thermodynamics and quantum theory. Initially, people studied the mechanics of black holes [1], and they found that if applying the quantum field theory in curved spacetime to black holes, they could get the profound results. The black hole thermodynamics is based on the discovery made by Bekenstein and Hawking that black holes have entropy and other thermodynamic quantities [2][3][4][5]. The black holes have negative specific heat in asymptotically flat spacetime which causes it to be an unstable system. The problem in asymptotically flat space then can be avoided since the gravitational potential of dS/AdS spacetime acts as a box of finite volume with unphysical perfectly reflecting walls [6]. The AdS-Schwarzschild black hole was investigated by Hawking and Page, where the phase transition from thermal AdS to black hole was found. By introducing the AdS/CFT correspondence [7] or the AdS/QCD correspondence [8,9], the phase structure of AdS black holes becomes more charming. Motivated by that fact, and the recent observations of the striking similarity between the thermodynamic phase structure of AdS black holes in the canonical ensemble and that of the van der Waals-Maxwell liquid-gas system, the physics of the charged AdS black holes were explored in more detail [10,11]. Regarding the negative cosmological constant as the thermodynamic pressure [12][13][14] enriches the phase structure of black holes [15][16][17][18][19][20][21]. In line with that, the phase transition and critical behavior of the black hole in the extended phase space have been numerously investigated in [22][23][24][25][26][27][28][29][30][31][32][33]. In another hand, there is a correspondence between a strong gravitational theory in de Sitter space and conformal field theory [34]. The thermodynamics study of a dS spacetime with black hole is interesting and important see Ref. [35]. That is due to the fact that the universe during the early-inflationary epoch was a dS spacetime, in far future it will turn into a dS one [36,37]. Furthermore, similar to the AdS/CFT correspondence in AdS spacetime with a black hole, the dS/CFT correspondence was apply to dS spacetime with a black hole in Ref. [38], in which more information about possible phase transitions in various dimensions was explored. Under these efforts, black holes in dS/AdS spacetime have been widely recognized as thermodynamic systems. What's more, the observation of black hole opens a new window for its thermodynamic research. While black holes in dS spacetime, in which the effect of black hole inner horizon is often neglected, cannot be in thermodynamic equilibrium in general due to the multiple horizons with different temperatures. To overcome this problem, one had analyzed one horizon and taken another one as the boundary or separated the two horizons by a thermally opaque membrane or box [39,40], in which the two horizons were analyzed independently. Besides, one could take a global view to construct the globally effective temperature and other effective thermodynamic quantities [41,42]. In this issue, there are three assumptions of constructing the total entropy in dS spactime with black hole: i) the sum of both horizons [43,44] with the volume V ef f = V c − V b . In this case the black hole and cosmology horizons are independent with each other; ii) the difference of both horizons [45,46] with V ef f = V c + V b + V in ; iii) the sum of both horizons and their interaction with [47][48][49]. The revelent works on the thermodynamic property of dS spacetime with black hole were presented [50][51][52][53][54][55][56][57][58][59][60]. Note that due to the existence of gravity in system, the interplay between the black hole and cosmology horizons should be considered, thus we adopt the third assumption in this work. In the other hand, for the space between the black hole outer horizon (generally called as the black hole horizon) and cosmology horizon, the different hawking temperatures on two horizons prevents a dS spacetime with black hole in equilibrium as like an ordinary thermodynamic system. While, there are common parameters M , Q, on the black hole and cosmological horizons. Thus, the thermodynamic quantities on two horizons are not independent. The interplay between two horizons must be taken into account when constructing the effective quantities of a dS spacetime with black hole in thermodynamic equilibrium. Based on these, the thermodynamic phase transition, Joule-Thomson expansion, and the continuous phase transition theory for the RN-dS spacetime will be further probed. In addition, despite the success of black hole thermodynamics, its microscopic origin is a mystery at present. The Ruppeiner geometry [61] derived from the laws of thermodynamic fluctuations gives us a glimpse of the microstructure of black holes. The Ruppeiner line element measures the distance between two adjacent thermodynamic fluctuation states, from which we can obtain the thermodynamic Riemannian curvature scalar. The corresponding curvature scalar marks the interaction between adjacent modules of fluid systems: a positive (negative) thermodynamic curvature scalar indicates that a repulsive (attractive) interaction domain and a noninteracting system such as the ideal gas corresponds to the flat Ruppeiner metric [62][63][64][65][66]. Thus, in this work, we will investigate microstructure of the RN-dS spacetime near the phase transition point by the Ruppeiner geometry. The work is organized as follows. In Sec. II, we would like to discuss the conditions for the existence of black hole and cosmological horizons in RN-dS spacetime, as well as the influence of spacetime parameters, and give the range of the position ratio x of two horizons. Then present the effective thermodynamic quantities and the critical point of the RN-dS spacetime, analyze the behaviors of the heat capacities, isobaric expansion coefficient, and isothermal compression coefficient. In Sec. III, based on the effective thermodynamic quantities of RN-dS spacetime, we extend the method of the phase transition for Van de Waals systems or charged AdS black holes, use it to study the RN-dS spacetime, and exhibit the first-order phase transition of the RN-dSQ spacetime. In Sec. IV, we explore the Joule-Thomson expansion of the RN-dS spacetime in isenthalphy processes. The inversion curves divide the isenthalpic ones into two parts in the P − T diagram: a cooling phenomena with the positive slope and a heating process with negative slope. In Sec. V, the critical exponents of the RN-dS spacetime are obtained and they are satisfied with the universal relations. In Sec. VI, we investigate the thermodynamic geometry of two coexistent phases for the RN-dS spacetime. The result shows the geometries of two coexistent phases are both positive, and the geometry of one phase is always bigger than the other phase. That means the RN-dS spacetime in two coexistent phases are domain with the repulsive interaction, and the average interaction of this system with one phase is bigger than the other phase. This phenomenon provides a new way for us to further study the microstructure interaction of dS spacetimes. The Sec. VII is the summary. II. REVIEW OF STABLE RN-DS SPACETIME We will review the solution of the RN black hole embedded in a dS spacetime. And we give the concrete process of exporting the effective thermodynamic quantities and the critical point of RN-dS spacetime in the induced phase space, and analyze the heat capacities, isobaric expansion coefficient, and isothermal compression coefficient, respectively. A. Solution of RN black hole in dS spacetime The static spherically symmetric solution of Einstein equation for a RN-dS black hole have been obtained with the horizon function where M and Q are the black hole mass and charge, l is the curvature radius of dS spacetime. There are three horizons: the RN-dS black hole inner horizon, the RN-dS black hole outer horizon (i.e., RN-dS black hole horizon), and the cosmological horizon. The radiuses of RN-dS black hole horizon and the cosmological horizon satisfy the expression f (r + , r c ) = 0. The mass parameter can be obtained from the above equation where M is the reduced mass of black hole. For similarity, we introduce the electric potential φ = Q/r, the metric function of (2) can be rewritten as On two horizons, the potentials read φ + = Q/r + and φ c = Q/r c . The reduced mass parameter becomes The plots of M − r and M − φ with the parameter Q 2 l 2 = 1 50 are given in Fig. 1. For the RN-dS spacetime, there exist the local minimum (M min ) and local maximum (M max ) for black hole mass. As M min ≤ M 0 ≤ M max , this system have three horizons, i.e., the black hole inner and outer horizons, the cosmological horizon. Note that the black hole horizon r + means the outer one. When the local minimum and local maximum merge into an inflexion point, these three horizons coincide together. In this case, the hole black is called the ultracold black hole (r ucold = √ 2Q, Q 2 3 ). Generally, the RN-dS black hole horizons do not coincide with the cosmological one, for a RN-dS system the conditions Q 2 l 2 ≤ 1 12 and M min ≥ M ucold should be held on. In the following we will investigate the thermodynamic property for this system from the view of the potential on horizon, instead of horizon radius. From eq. (5) and the definition x ≡ φc φ+ , we have When M 0 = M min > M ucold : the black hole inner and outer horizons coincide together, such black hole is called the cold one. As M 0 = M max > M ucold : the black hole and cosmologic horizons coincide together (x = 1), this black hole is called the Nariai one whose maximum potential becomes Generally, the potential on horizon is a limited value, not infinity, i.e., the ratio of two horizon potentials should be from a minimum instead of zero to one. For the Nariai black hole, substituting eq. (7) into eq. (6), the minimum of x is constrained by the following expression with Q 2 l 2 ≤ 1 12 , whose picture is shown in Fig. 2. In this system, the RN-dS black hole can be survives with the range of x min ≤ x ≤ 1. B. Effective thermodynamic quantities of RN-dS Spacetime For a dS spactime with RN black hole, one can describe the thermodynamical property on the black hole and cosmological horizons, respectively, where the thermodynamical laws are still satisfied. Since the first law are held on for the black hole and cosmological horizons, the thermodynamic quantities on two horizons are When the hawking radiation temperatures on two horizons are equal to each other, i.e., in the lukewarm case [35] the potentials on two horizons become and the same radiation temperature on two horizons reads Since there are two different Harking-temperature on the RN black hole and cosmological horizons, we can not directly regard the space between the black hole and cosmological horizons as an ordinary thermodynamic system in equilibrium to study the phase transition. While considering the gravity effect between two horizons' space, i.e., the interplay of two horizons should be introduced, the RN-dS spacetime with the space between two horizons can be treated as a thermodynamic system affected by gravitational effects, where the thermodynamical laws are still held on. From the view of the whole RN-dS spacetime, two horizons are not independent with each other, because they are in gravity field which will leads to the interaction between them. For the RN-dS spacetime, we mainly investigate the space between black hole and cosmological horizons, i.e., the thermodynamic volume of our considered system is The boundary of our considered system are two horizons which have different radiation temperatures. That is because the system is in gravity field, and the horizon temperatures are deformed by the warped space. Thus we regard this system as an ordinary thermodynamic system in the thermodynamic equilibrium, which is of the thermodynamic quantities ( T ef f , P ef f , V , S or T ef f , P ef f , V, S). Next we will investigate the thermodynamical quantities of RN-dS system: such as T ef f , P ef f , V, S, which are called the induced thermodynamical quantities and they are both the functions of φ + and x. Here we point out the entropy is not only the sum of two horizons, it should contain the connection between two horizons which arise from the gravity effect. We assume the total entropy of this system as the following form where f 0 (x) stands for the interplay between two horizons and it arises from the gravity effect. In the following, we mainly focus on the process of obtaining the function f 0 (x). When regarding the RN-dS spacetime as an ordinary thermodynamic system in the thermodynamic equilibrium, the first law should be held on Thus the effective temperature and pressure can be obtained from the following expressions When Q 2 l 2 satisfies eq. (6), the temperature corresponding to two horizons is equal to each other. In this case, we believe that the effective temperature should be the radiation one As x = 0, the interplay between two horizons should be vanished, i.e., f 0 (0) = 0. By solving above equation the interaction function of entropy reads Substituting eq. (19) into eqs. (16) and (17), the effective temperature and effective pressure are satisfied the following forms as shown in Ref. [58] In order to understand the temperatures on two horizons and the effective temperature with different Q 2 /l 2 , we present the picture of them in Fig. 3. Adopting Q, S, V as the independent parameters of M and considering the units of these quantities, the smarr formula reads In the same way for the reduced quantities T ef f , P ef f , S, V, M , the reduced smarr formula becomes These results can be verified by substituting eqs. (20), (14), (13), (21), and (36) into the above equation. The critical point denoting a second-order phase transition is determined by the following equations With eqs. (13), (19), (20), and (21), the critical reduced thermodynamical quantities are Performing isobaric, isothermal, and isovolume processes for the system and solving eqs. (20), (21), and (13), the physical potentials on RN-dS black hole horizon φ + become For a real RN black hole embedded in a dS spacetime, since the black hole horizon is not coincide with the cosmological horizon, the value of x cannot be one. In order to solving this problem, we study the heat capacity at the constant Q 2 /l 2 and at the constant effective pressure P ef f , respectively, i.e., for a real stable RN-dS spacetime the heat capacities should be positive. The heat capacities at the constant Q 2 /l 2 and P ef f are Here the reduced heat capacities read Combining eqs. (6), (14), (19), (20) and (26), the pictures of heat capacities with different values of Q 2 /l 2 and P ef f are shown in Fig. 4. As what we have mentioned, from Fig. 4(a) we can see that the value of x for a stable RN-dS spacetime is from x min to x max , where the maximum and minimum of x is related with Q 2 /l 2 . Note that for the given Q 2 /l 2 the minimum of x in eq. (29) is bigger than that in eq. (8). Therefore, the thermodynamical property of RN-dS spacetime will be investigated in a certain range of x obtained from the view point of C Q 2 /l 2 . In Fig. 4(b), C P ef f is divergent at the critical point and at two different values of x when P ef f < P c ef f . In the next, we will present the two divergent location of C P ef f when P ef f < P c ef f are just the ratios between two horizons' potentials of two coexistent phases. Similarly, the reduced heat capacity with the constant volume becomes From eqs. (13), (14), (19), and (20), the plots of the heat capacity with different volumes are exhibited in Fig. 5. In addition there are some coefficients to probe the critical behavior of a thermodynamical system, i.e., the isobaric expansion coefficient and isothermal compression coefficient The introduced isobaric expansion coefficient and isothermal compression coefficient have the following forms Their behaviors nearby the critical point are exhibited in Fig. 6. III. FIRST-ORDER PHASE TRANSITION In this section, we would like to study the first-order phase transition in the canonical ensemble. From the effective thermodynamical quantities of RN-dS spacetime, the thermodynamical state equation can be written as Gibbs free energy of this system reads with the reduced mass parameter Combining eqs. (13), (14), (19), (20), (21), (26), and (27), the behaviors of Gibbs free energy nearby the critical point in 2-dimension and 3-dimension are shown in Fig. 7. In addition, the first-order phase transition point can also be obtained by the construction of equal-area law in different phase diagrams. Here we exhibit the first-order phase diagrams of P ef f − V and T ef f − S in Fig. 8. In fact, the first-order phase transition point is a coexistent state of two different phases in RN-dS spacetime. All these points form one coexistent curve. Although there is no the analytic form for the curve, it can be described the following behavior with the numerical results in Fig. 9. And the thermodynamic quantities of two phases with different first-order phase transition points are presented in Tab. I. Now we exploit Ehrenfest's scheme in order to understand the nature of the phase transition, which basically consists of a pair of equations known as Ehrenfest's equations and second kind. For a standard thermodynamic system these equations may be written as [67][68][69] For a genuine second-order phase transition both of these equations have to be satisfied simultaneously. With the expressions , the Prigogine-Defay (PD) ratio [70] of the RN-dS spacetime may be found to be Hence the phase transition occurring at T c ef f is a second-order equilibrium transition, which is similar to that of AdS black hole. This is true in spite of the fact that the phase transition curves are smeared and divergent near the critical point. IV. JOULE-THOMSON EXPANSION More recently, the authors of [71] have investigated the Joule-Thomson (JT) expansion for AdS charged black holes with the aim to confront the resulting features with those of Van der Waals fluids. The extension to rotating-AdS black hole [72] and the charged black hole solution in the presence of the quintessence field [73] have also been considered. JT expansion [74] is a convenient isoenthalpic tool that a thermal system exhibits with a thermal expansion. It is worth noting that when expanding a thermal system with a temperature T , the pressure always decreases yielding a negative sign to ∂P . In this section, we will investigate the Joule-Thomson expansion of the RN-dS spacetime. In Joule-Thomson expansion for the Van der Waals system as well as AdS black holes, gas/black hole phase is passed at high pressure through a porous plug or small value in the low-pressure section of an adiabatic tube, and the enthalpy remains constant during the expansion. The expansion is characterized by a change in temperature relative to pressure. In addition the enthalphy can be used to define the non-equilibrium states. Thus we can also introduce it to describe the RN-dS spacetime. The Joule-Thomson coefficient, which can describe the expansion process, is read as where the enthalphy is For similarity, we introduce the reduced enthalphy and the Joule-Thomson coefficient as For an isenthalphy for this system, the relation between potential and two horizons' ratio x reads with When αT ef f = 1, i.e., µ = 0, we can obtain with To better investigate the Joule-Thomson expansion, by substituting eqs. (44) and (46) into eqs. (20) and (21) respectively, the isoenthalpic and inversion curves of this system are depicted in Fig. 10. The result shows that the inversion curve divide the isoenthalpic one into two parts: one is the cooling phenomena with the positive slope of the P − T curves, the other is the heating process with the negative slope of the P − T curves. V. LANDAU' THEORY OF CONTINUOUS PHASE TRANSITION For an ordinary thermodynamic system, the second-order derivative of Gibbs free energy, heat capacity, isobaric expansion and isothermal compression coefficients have a peak or emerge a jump nearby the critical point. Additionally, the power functions also are used to describe the critical behavior, whose powers are called the critical exponents. • Spontaneous magnetization satisfies M ′ ∝ (−t) β ′ ,t → 0 − , and it equals to zero when above the critical temperature; • Magnetic susceptibility obeys χ ∝t −γ ′ ,t → 0 + ; χ ∝t −γ ′′ ,t → 0 − ; • Magnetization and applied magnetic field satisfies M ′ ∝ H 1/δ ,t → 0; • Heat capacity without applied magnetic field of ferromagnetic matters obey the following expressions Subsequently, people found that for various fluid systems the critical behavior is the same as well as ferromagnetic systems. Furthermore, the critical behavior of systems with completely different physical characteristics have extremely similar characteristics compared with fluid and ferromagnetic systems. That indicates the critical phenomena has a certain universality. Based on the investigation of AdS black holes, the critical powers calculated theoretically are consistent with that of ferromagnetic systems. For the RN-dS spacetime, the critical phenomena should be also focused. However, it is different from AdS black holes and ferromagnetic systems, and its phase transition is determined by the ratio of electric potentials between two horizons as shown in Sec. IV. Hence, the explore of critical phenomena for RN-dS spacetime will help us better understand its thermodynamic properties. For the Landau' theory of continuous phase transition, the phase transition is accompanied by a change in the degree of order and symmetry of system. By regarding the RN-dS spacetime as a thermodynamic system which is composed of internal molecules, whether there also exist the changes of internal symmetry and order when phase transition emerges becomes a question of general concern. Though the investigation shown in Sec. III, we found that when the phase transition emerges, for the system in the higher-potential black hole phase, the internal molecules have a certain orientation under the action of strong electric potential, and are in an orderly and high symmetric state; while for the system in the low-potential black hole phase they are in a low-order and low symmetric state. Increasing temperature above the critical one, the thermal motion of internal molecules increases, making their orientation tend to zero. In order to explain the nature of phase transition of RN-dS spacetime, a new order parameter η related to electric potential should be introduced to describe the degree of order. When T ef f < T c ef f , the electric potentials on RN-dS black hole and cosmological horizons both have two different values: φ +1,+2 and φ c1,c2 . With the definition y ≡ φ +1 /φ +2 , the difference of the electric potentials on black hole horizon for two black hole phases becomes △φ + = φ +1 (1 − y). As T ef f → T c ef f , y → 1, η → 0. Since △φ + = 0 with T ef f ≥ T c ef f and △φ + = 0 with T ef f ≤ T c ef f , △φ + can be adopted as the order parameter: η ≡ △φ + . Because the order parameter is a small quantity nearby the critical point and the system is symmetry for η ⇆ −η, for the simplicity the Gibbs free energy G(T ef f , P ef f ) can be extended to the fourth order in the power series of the order parameter as following form The corresponding generalized force ξ reads Hence the critical exponent δ = 3. For the system in the full disorder state: ξ = 0, by solving the above equation we have By analyzing the above solutions we can find that when B > 0 the system will be in a stable disorder state with η = 0, A > 0; while it is in a stable order state with η = ± − A 2B , A < 0. The exponent β ′ equals 1/2. Because of A(T c ef f , P c ef f ) = 0, we can assume with the positive coefficients A 1 (T c ef f , P c ef f ) and A 2 (T c ef f , P c ef f ). For T ef f > T c ef f and P ef f = P c ef f , the RN-dS spacetime is in a stable disorder state, otherwise it is in a stable order one when T ef f < T c ef f and P ef f = P c ef f . Nearby the critical point, the Gibbs free energy, entropy, and volume are continuous and have the following forms approximately . The heat capacities and isothermal compression coefficient at the critical point are T c ef f , which indicate that they both have a jump or a peak at critical point as well as the ferromagnetic systems. The critical exponents are α ′ = α ′′ = 0. Since χ −1 = ∂ 2 G ∂η 2 = 2A + 12Bη 2 when P ef f = P c ef f , we have and the critical exponents are γ = γ ′ = 1. Furthermore, for the RN-dS spacetime there exist the same scale relations VI. THERMODYNAMIC GEOMETRY OF RN-DS SPACETIME As shown in the last section, the critical exponents are all independent with A and B, as well as an ordinary thermodynamic system. The reason is that the fluctuation of the order parameter η near the critical point is neglected. In Ref. [62], the authors investigated the phase transition structure of black holes through the singularity of the spacetime scalar curvature R. The scalar curvature is divergent at the critical point of AdS black hole phase transition as fluid systems do. This property can be used to investigate the information contained in RN-dS spacetime phase transition from the thermodynamic geometric perspective. The sign of R can be used as a diagnostic of the interaction between two microscopic constituents (or molecules) of this system. A positive scalar curvature indicates a repulsive interaction, whereas a negative one stands for an attractive interaction. In this part, we will investigate the scalar curvature to reveal the microstructure of this system. For simplicity, we adopt the fluctuation coordinates T ef f and V , the metric is two-dimensional and diagonal Since the heat capacity C V = T ef f ∂S ∂T ef f V , the line element can be expressed as the scalar curvature is easily worked out Combining eqs. (13), (20), (21), (31), and (55), the behaviours of the scalar curvature for two different phases with the horizons' ratio x is displayed in Fig. 11. From Fig. 11, we find that the values of scalar curvature for two-coexistent phases are both positive, which means the average interaction of the system in two different coexistent phases is repulsive. Since R 2 ≥ R 1 , the interaction of the phase φ +2 is bigger than the one with the phase φ +1 , until they are equal to each other at the critical point. VII. CONCLUSION In this work, when considering the gravity effect and the interplay between the black hole and cosmological horizons of the four-dimensional RN-dS spacetime, we presented the effective thermodynamic quantities of RN-dS spacetime in thermodynamic equilibrium from the viewpoint of the electric potential on black hole horizon (φ + ). Based on these effective quantities, the thermodynamical properties of RN-dS spacetime were investigated. This provides a good way to further understand the relationship between the gravity and the thermodynamics of black holes. First, with a certain set of parameters we gave the condition of the ratio between the black hole and the cosmological horizons which were both survived in the four-dimensional RN-dS spacetime. And the concrete process of exporting the effective thermodynamic quantities and critical point of RN-dS spacetime were also presented. Through the analyse of the heat capacities, the isobaric expansion coefficient, and the isothermal compression coefficient, respectively, we found that there exist a peak behavior both of these quantities which are as the function of x. Those results indicate the existence of the second-order phase transition of this system in the certain parameters. Then, the first-order phase transition in the induced phase space was also exhibited by two methods: the Maxwell's equal area law and Gibbs free energy, which is independent on the model parameters. And the two-phase coexistent curve in the P − T plane was presented. For this system, the Prigogine-Defay (PD) ratio equals to one which is consistent with that in the ordinary thermodynamical systems as well as AdS black holes. In order to describe the thermal expansion with an isenthalphy process, the properties of the isoenthalpy and inversion curves for this system were also shown. The isoenthalpic curve is divided into two parts by the corresponding inversion curve: the cooling phenomena with the positive slope and the heating process with the negative slope of the P − T curve. In addition, based on the Landau's continuous theory of phase transition we presented the critical exponents and the scale relations, which are consistent with AdS black holes as well as the ferromagnetic systems. Those similarities among in the RN-dS spacetime, AdS black holes, and the VdW systems imply that the RN-dS spacetime and AdS black holes should be of the internal microstructure as well as the VdW system. Finally, to probe the microstructure of the RN-dS spacetime we introduced the scalar curvatures of two coexistent phases by the Ruppeiner geometry method that is used in AdS black hole thermodynamics. The results indicate the average interactions of the system in two different coexistent phases are repulsive. Furthermore the interaction of the high-potential phase φ 2 is bigger than the one with the low-potential phase φ 1 , until they are equal to each other at the critical point.	2022-07-08T01:16:12.732Z	2022-07-07T00:00:00.000	{ "year": 2022, "sha1": "122ebdddda04db38e786f4354813972eaed7ea20", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1140/epjc/s10052-022-10797-w.pdf", "oa_status": "GOLD", "pdf_src": "Arxiv", "pdf_hash": "122ebdddda04db38e786f4354813972eaed7ea20", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
15011447	pes2o/s2orc	v3-fos-license	Blinding Orbital Apex Syndrome due to Onodi Cell Mucocele The onodi cell is a posterior ethmoidal cell that is pneumatized laterally or superiorly to the sphenoid sinus with close proximity to the optic nerve. A mucocele, a benign, expansile, cyst-like lesion of the paranasal sinuses, may uncommonly involve the onodi cell causing compression of the optic nerve and nearby structures. In this paper, we report a rare case of onodi cell mucocele causing orbital apex syndrome, with prompt recovery after endoscopic removal. However, optic neuropathy did not improve and the patient remained blind. Introduction Mucoceles arising from the sphenoid sinus or a posterior ethmoidal cell are uncommon and may result in compression of the optic nerve and adjacent structures, leading to orbital apex syndrome. Onodi cell is a posteriorly positioned ethmoid cell that pneumatizes laterally or above the sphenoid sinus, adjacent to the course of the optic nerve. Mucocele formation in such a cell may lead to compressive optic neuropathy and blindness, as in the case presented here. Case Report A 53-year-old female presented with vision loss in her right eye that started one day earlier, accompanied by pain in eye movement. On examination, visual acuity was finger counting in her right eye and 20/20 in her left eye. There were also swelling and erythema of the right eyelids and a right relative afferent pupillary defect. The rest of the ophthalmic examination was normal. She was afebrile with no signs of active infection. CT demonstrated opacification of the ethmoidal and sphenoidal air cells parallel to the optic nerve ( Figure 1). Patient history was positive for chronic rhinosinusitis that was treated one year earlier with an uneventful bilateral total functional endoscopic sinus surgery (FESS) that included the sphenoid sinuses. With a diagnosis of an inflammatory process, treatment with intravenous methylprednisolone (1 mg/kg per day) was started. A few hours later, the patient complained of binocular diplopia and paresthesias along the distribution of cranial nerves V1 and V2. In addition, eyelid swelling had worsened and visual acuity deteriorated to light perception only. An almost complete limitation of abduction and adduction in her right eye was noted ( Figure 2). An urgent MRI demonstrated an onodi cell eroding through the lamina papyracea, extending to the superior orbital fissure and compressing the optic nerve, the first 2 divisions of the trigeminal nerve, and the abducens nerve. On the same day the patient underwent an endoscopic sinus surgery with cyst marsupialization and aspiration of a clear mucoid substance. Cultures were not taken due to the clinical observation of a mucocele. During the following days eye movements returned to normal but right optic disc atrophy evolved with visual acuity of no light perception. Discussion There are only a few reports on onodi cell mucocele causing orbital apex syndrome [1][2][3][4]. Paranasal sinus mucoceles affect the optic nerve by either mechanical compression causing ischemic injury or by direct extension of the inflammatory process to the optic nerve [5]. It has been suggested that cases with gradual visual acuity decrease are caused by circulatory disorders due to direct pressure of the mucocele, whereas cases that show rapid loss of visual acuity are caused by direct spread of infection or inflammation to the optic nerve [5]. Our patient demonstrated rapid visual loss. When inflammatory signs subsided and no recovery of the optic nerve function was noted, it was suspected that ischemia of the posterior optic nerve may have occurred. The finding of clear mucoid fluid also supports the diagnosis of ischemic injury to the optic nerve. However, considering the acute inflammatory signs, inflammation may also have a role in the presence of optic neuropathy. Treatment of mucocele is based on prompt endoscopic surgical decompression and systemic steroids to reduce inflammation [1]. When treated early, visual outcome may improve dramatically [1,2]. Unfortunately the patient's poor prognostic factors, such as sudden onset and poor visual acuity on presentation, led to development of optic atrophy and loss of vision. In conclusion, clinical signs of orbital apex syndrome in patients with a history of chronic sinusitis should raise the suspicion of mucocele arising from the posterior paranasal sinuses. An immediate orbital and sinuses imaging is required with early surgical intervention. With prompt treatment, vision and eye movements can be restored even though permanent visual loss may occur.	2018-04-03T05:58:07.224Z	2014-05-15T00:00:00.000	{ "year": 2014, "sha1": "3aaef8ff3147ab270ac01e37272eed9543d341ab", "oa_license": "CCBY", "oa_url": "http://downloads.hindawi.com/journals/criopm/2014/453789.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "1387bce67684b7bdb8bee277c8c21852efa9471d", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
211190039	pes2o/s2orc	v3-fos-license	Integral use of Argentinean Solanum betaceum red fruits as functional food ingredient to prevent metabolic syndrome: effect of in vitro simulated gastroduodenal digestion The “red chilto” (Solanum betaceum Cav) is a native fruit from the Yungas forest of Argentina. Red chilto is a neglected and underutilized native species (NUS). The objective of this work was to evaluate the potentiality of pulp, seed and skin from “red chilto” as a functional food ingredient to add value to a native resource of Argentine Yungas to promote sustainable integral use of it. The powders have low carbohydrate and sodium content and are a source of vitamin C, phenolic acids (rosmarinic acid and caffeoylquinic acid), anthocyanins, condensed tannins, carotenoids, potassium, and fiber. The phenolics of chilto powders showed, before and after simulated gastroduodenal digestion, antioxidant activity (ABTS•+; H2O2; O2•; HO•) and were able to inhibit enzymes related to metabolic syndrome, such as α-glucosidase, α-amylase and lipase. Chilto powder showed hypoglycemic effect by increasing glucose adsorption, decreasing glucose diffusion rate and by promoting glucose transport across the cell membrane. These results suggest the potential of Argentinean “red chilto” fruits as functional food ingredients or dietary supplements to prevent metabolic syndrome principally by its antioxidant, hypoglycemic and hypolipemic effects. Introduction The study of native fruits as unconventional foods is an emerging area in Argentina (Cardozo et al., 2011;Costamagna et al., 2013;Orqueda et al., 2017). In Northwestern Argentina, many native fruits make up the food spectrum of rural communities. Solanum betaceum Cav., called "monte tomato" or "chilto" is one of them. S. betaceum is distributed in Northwest Argentina in the provinces of Catamarca, Salta, Jujuy and Tucum an in the "Selva Pedemontana" of the Yungas forest, between 400 and 700 m.a.s.l. It plays an outstanding ecological role in the context of the Yungas since it has a high biological diversity and serves as a refuge for species of other altitudinal levels of the mountain forest. The Selva Pedemontana has been strongly reduced in size by the transformation of its natural cover in areas of intensive agriculture, in particular, sugar cane. S. betaceum produces edible fruits with an interesting growing market in its native Andean countries, as well as in other parts of the world (Prohens and Nuez, 2000). The main fruit types were conventionally recognized based on skin colour, namely dark purple, red and orange. At present, the sustainable crop of red and orange chilto is being carried out in Argentina Yungas in three provinces. The orange variety is being grown in Tucum an province and the red variety in Jujuy and Salta provinces. The orange variety of chilto was previously investigated for nutritional and phytochemical composition, antioxidant and inhibitory activity on key enzymes involved in metabolic syndrome (Orqueda et al., 2017). However, so far there have been no studies of the red variety grown in Argentina. Although the native fruits can be incorporated in the diet as fresh ripe fruits, they have a limited shelf life, as physical, chemical and biological changes can occur after a few days of harvest. Under these circumstances, minimum processing is often necessary to achieve a longer shelf life, preserving their nutritional and functional values as well as their organoleptic properties. The objective of this work was to evaluate the potentiality of pulp, seed and skin from "red chilto" as a functional food ingredient to add value to a native resource of Argentine Yungas and promote sustainable integral use of it. Plant material The fruits of Solanum betaceum (red variety) were harvested at the ripening stage in which they are consumed in Finca del Obispo, Villa Jardín de Reyes, Jujuy, Argentina, in December 2016. Skin, pulp and seeds were separated, freeze-dried and powdered. The powders were vacuum packed and stored at -20 C. Scanning electron microscopy (SEM) of each powder was carried out. SEM of gold-coated (Fine Coat Ion Sputter JEOL JFC-1100) samples was performed by using a ZEISS SUPRA-55 VP field emission scanning electron microscope at Centro Integral de Microscopía Electr onica (CIME), CONICET-UNT. The micrographs were taken for each sample at a magnification of 10,000 and 50,000 ( Figure 1). Quality parameters of fresh fruits The titratable acidity was analyzed for each sample (AOAC, 2000). To determine the different color parameters, a Chroma meter CR-400 colorimeter (Konica Minolta, Tokyo, Japan) was used through the CIE-Lab system. Chromaticity coordinates L , a and b * were obtained. The parameters were obtained from three independent measurements in the equatorial region of each whole fruit. Ten fruits were measured in total. Determination of metabolic profile of fresh fruits by 1 H-NMR To test the metabolic profile, whole fresh fruits were frozen in liquid nitrogen. Then, was homogenized with potassium phosphate buffer (1 M pH 7.4 in D 2 O) and centrifuged at 4 C for 30 min at 14000 rpm. The supernatant was recovered and the pH was adjusted to 7.4. The deuterated sodium salt of 3-trimethylsilyl-1-propane sulfonic acid (TSP) (except for the methyl groups bound to the silicium atom) was added as an internal standard. 1 HNMR at 600.13 (MHz Bruker Avance II spectrometer) was used. Calculation of sweetness index The sweetness index for each powder was determined according to Keutgen and Pawelzik (2007). The sweetness index of each part of the chilto fruit, an estimate of total sweetness perception, was calculated based on the amount and sweetness properties of individual carbohydrates presents in the samples. The contribution of each sugar in the samples, taking into account that fructose and glucose are 1.7 and 0.75 times sweeter than sucrose, respectively was estimated. As a result, sweetness index ¼ 0.75 [Glucose] þ 1.0 [Sucrose] þ 1.7 [Fructose]. Estimation of taste parameter of the fruit powder The sugars were repeatedly extracted from each part of the fruit using 80% ethanol at 80 C until their total depletion. Total soluble sugars were quantified by the Fenol-H 2 SO 4 method proposed by Dubois et al. (1956). Reducing sugars were quantified using the Somogyi-Nelson method (Nelson, 1944;Somogyi, 1945). The glucose content was determined using the enzyme glycemia kit (Wiener lab. 1400101) which is based on the glucose oxidase/peroxidase method. Fructose and sucrose were determined using the resorcinol-thiourea method proposed by Cardini et al. (1955) and Roe and Papadopoulus (1954). 2.6. Proximal composition of chilto powder 2.6.1. Protein, fat, fiber and minerals To determine of crude protein, lipids and minerals content, the protocols previously described by Orqueda et al. (2017) was used. The total protein content was determined by measuring the total nitrogen (N) by the Kjeldahl method with a conversion factor of 6.25 (AOCS, 1989). For the determination of the total lipid content, the samples were extracted by Soxhlet with petroleum ether (40-60 C) for 4 h and gravimetrically quantified. The fiber content was determined according to AOAC (2005). Briefly, 1 g of fat-free powder was weighed, treated with 20 mL of 1.25% H 2 SO 4 and incubated at 100 C for 30 min. Then, 50 mL of 3.52% NaOH was added and incubated again at 100 C for 30 min. Finally, the sample was dried at 110 C until constant weight. It was then incinerated in muffle at 500 C and the crude fiber content was determined by weight difference. The integrated total dietary fiber kit (WI 96461, Megazyme International Ireland, Ireland) was used to determine the dietary fiber content (McCleary, 2014). Mineral analysis was performed using quadrupole induction plasma mass spectrometry (Q-ICPMS) at the Instituto Superior de Investigaci on, Desarrollo y Servicios en Alimentos, ISIDSA, C ordoba, Argentina. The following ions were analyzed: sodium, magnesium, potassium, calcium, iron, copper and zinc. Starch determination The extraction and determination of the digestible and resistant starch content were carried out according to the method of Holm et al. (1986) with some modifications. The powders (0.1 g) were mixed with 10 mL of 0.1 M HCl and 0.1 g of pepsin at pH 1.5 and kept at 40 C for 1 h. Then, 0.1 g of sodium azide was added and the mixture was hydrolyzed using 0.5 g of α-amylase (10 U/mg) at pH 6.9 for 16 h at 37 C. Subsequently, the samples were centrifuged at 6,000 x g. The pellets were dispersed in 2 M KOH and maintained at room temperature for 30 min. Then, the digestible starch and the resistant starch were subjected to a digestion process using 50 μL of a solution of amyloglucosidase 10 mg/mL (14U/mg) pH 4.75 in a water bath for 30 min at 60 C. The glucose content was determined by the glucose oxidase enzymatic assay (Wiener Lab., Argentina). Extraction and determination of phenolic compounds The phenolic compounds were obtained from chilto powders at 25 C with 95 ethanol (1:5, W/V) until exhaustion with ultrasound during 30 min (Pothitirat et al., 2009). Then, the samples were filtered and dried under reduced pressure to obtain the phenolic enriched extract (PEE). In each PEE, total phenolic content and flavonoid content was determined according to Costamagna et al. (2013). Extraction and quantification of tannins and anthocyanins were carried out according to Costamagna et al. (2013). Vitamin extraction and determination For vitamin C quantification, the powder was extracted with 2% H 3 PO 4 ultrasound assisted during 10 min and the ascorbic acid content was determined according to Costamagna et al. (2013). Vitamin C was expressed as g L-ascorbic acid (g L-AA)/100 g powder. Phenolic profiling by HPLC-DAD The extracts of the chilto samples (peel, pulp, and seeds) were analyzed by HPLC-DAD to obtain a phenolic fingerprint. The HPLC system used was a Shimadzu equipment (Shimadzu Corporation, Kyoto, Japan) consisting of an LC-20AT pump, an SPD-M20A diode array detector, CTO-20AC column oven and a LabSolution software. An Inertsil ® ODS-3 RP-C18 (5μ, 250 Â 4.6 mm) column (GL-Sciences Inc, Tokio, Japan) maintained at 25 C. The analyses were performed using a linear gradient solvent system consisting of 1% formic acid in water (A) and acetonitrile (B) as follow: 90%-85% A over 15 min, maintained to 20 min, then 85% A for 20-25 min, 85% to 70% A for 25-50 min, maintained to 70% A for 50-80 min, and returning to 90% A for 80.01-90 min. The flow rate was 0.5 mL/min and the volume injected was 20 μL. The compounds were monitored at 254, 280, 320 and 350 nm and spectra from 200 to 600 nm were recorded. The main phenolics were quantified. The reference compounds rosmarinic acid (98%, Sigma Aldrich, St. Louis, MO, USA) and 3-caffeoylquinic acid (98%, Phytolab GmbH, Verstenbergsgreuth, Germany) were used in concentrations ranging from 5-500 mg/L. Results were expressed as mg of the corresponding compound/100g extract. Simulated gastroduodenal digestion The chilto powder was submitted to a simulated gastroduodenal digestion process (GD) according to Orqueda et al. (2017) using 0.5 g of each sample (skin, pulp and seed powder). Then, the powder samples were centrifuged at 10000 x g and the digested powders and without digestion were used to determine adsorption capacity and diffusion of sugars. The polyphenolic enriched extracts (PEE) were also submitted to GD according to Orqueda et al. (2017). After GD, the polyphenols were extracted from the aqueous phase using ethyl acetate, dried and resuspended in dimethyl sulfoxide (DMSO) to enzymatic inhibition assays and in ethanol 95 to antioxidant assays. In all case, a negative control was performed using each of solvent. 2.9. Inhibition of enzyme involved in metabolic syndrome by chilto PEE 2.9.1. α-Glucosidase inhibition The α-glucosidase inhibition activity of PEE was tested according to Costamagna et al. (2016). The α-glucosidase enzyme and PEE of each part of the fruit (2.5-20 μg/mL) (before and after simulated GD digestion) were pre-incubated at 4 C for 10 min. To start the reaction, the substrate p-Nitrophenyl-α-D-glucopyranoside was added. The results were expressed as IC 50 values (μg GAE/mL of PEE necessary to inhibit the 50% of enzyme activity). Orlistat (tetrahydrolipstatin, ATC code: A08AB01, Elea Laboratory, Ciudad Aut onoma de Buenos Aires, Argentina) was used as a positive control. α-Amylase inhibition The inhibitory activity of α-amylase was measured using Amilokit (Wiener Lab Group, Rosario, Argentina, Cat. N 1021001) as reported by Costamagna et al. (2016) and following the manufacturer's instructions. Different concentrations of extract were used (5-40 μg GAE/mL) (before and after the simulated GD) and the results were reported as IC 50 values. Acarbose was used as a positive control. Lipase inhibition The effect on lipase activity was tested by measuring of enzymatic hydrolysis of p-nitrophenyl palmitate according to Costamagna et al. (2016). The assays were carried out in presence (final concentration between 2.5 and 20 μg/mL) and in absence of each PEE, before and after simulated GD digestion. After pre-incubation (enzyme-PEE) at 4 C for 10 min, the enzyme reaction was carried out for 20 min. IC 50 values were calculated. Orlistat was used as a positive control. The glucose adsorption capacity of each powder (before and after GD digestion) was determined by Ahmed et al. (2011). The reaction mixture consisted of 10 mL of 20 mM glucose solution and 0.5 g of powder from each part of the fruit. The mixture was stirred well, incubated in a water bath with a stirrer at 37 C for 4 h, centrifuged at 4,000 x g for 20 min and the glucose content in the supernatant was determined. The glucose bound to the powder was calculated using the following formula: Effect of chilto powder on glucose diffusion The assay was performed according to Ahmed et al. (2011). To carry out the analysis, 5 mL of a 20 mM glucose solution and chilto powder samples (before and after GD digestion) (0.25, 0.5 and 1g) were dialyzed in dialysis bags against 40 mL of distilled water at 37 C. The glucose content in the dialysate was measured at 30, 60, 120, 180 and 240 min using the enzymatic glycemia kit. A control was carried out without powder. Effect of PEE on glucose intake by Saccharomyces cerevisiae cells The assay was carried out according to Bhutkar et al. (2017). To prepare the suspension, commercial yeast was washed three times with distilled water and then centrifuged (3000 x g, 5 min) until the supernatant was clear. Then, a 10% (v/v) S. cerevisiae cells suspension was prepared in distilled water. One hundred microliter of this suspension was contacted with increasing concentrations of PEE (10-100 mg/mL) (before and after simulated GD digestion). Then, 1mL of 20 mM glucose solution was added to this mixture, incubated for 1 h at 37 C and centrifuged at 2500 x g for 5 min. The glucose concentration in the supernatant was determined by enzymatic glycemia kit. The percentage of increase in glucose consumption by S. cerevisiae cells was calculated using the following formula: For scanning electron microscopy, yeasts with and without treatment were coated with gold (Fine Coat Ion Sputter JEOL JFC-1100). ZEISS SUPRA-55 VP field emission scanning electron microscope was used. The micrographs were taken for each sample at a magnification of 10,000 and 50,000. Antioxidant activity of PEE Unless otherwise specified, antioxidant activity tests were carried out in 96 multiwell plates. Total antioxidant capacity assay The total antioxidant activity of PEE was tested using the ABTS radical-cation method (ABTS þ ) (before and after simulated GD digestion) (Orqueda et al., 2017). Results were expressed as SC 50 (the concentration of PEE necessary to scavenge 50% of ABTS). BHT and quercetin were used as reference compounds. Protection of oxidative hemolysis assay The capacity of PEE to protect the red blood cell (RBC) membrane from oxidative hemolysis was carried out according to Orqueda et al. (2017), before and after the simulated GD digestion (0.1-2.5 mg GAE/mL). Antioxidant activity was measured using 2 mL cuvettes. The IC 50 values were determined as the concentration of PEE necessary to protect the RBC from oxidative hemolysis by 50%. Quercetin was used as a reference compound. H 2 O 2 scavenging assay The hydrogen peroxide scavenging assay was assayed according to Fernando and Soysa (2015) method with a few modifications. PEE, before and after simulated GD digestion (2.5-40 mg GAE/mL) and H 2 O 2 were pre-incubated for 3 min at 37 C. Then, a solution of phenol (12 mM) and 4-aminoantipyrine was added to the mixture. The content of hydrogen peroxide was determined catalyzing its conversion by horseradish peroxidase (HRP) and measured spectrophotometrically by the formation of a coloured quinone at 504 nm. The results were expressed as SC 50 values (mg GAE/mL), the concentration necessary to scavenge 50% of H 2 O 2 . Quercetin and ascorbic acid were used as controls. Superoxide radical scavenging assay The capacity of PEE to scavenging superoxide radicals was measured with the phenazine methosulfate (PMS) method (Cardozo et al., 2011). The superoxide radicals were generated by oxidation of NADH (β-nicotinamide adenine dinucleotide) in a system with PMS and were measured spectrophotometrically by the reduction of nitroblue tetrazolium (NBT). The reaction mixture contained extract, before and after simulated GD digestion at 10-100 mg GAE/mL. Then, the reaction was incubated 20 min at 25 C and the optical density was read at 560 nm. The SC 50 values were measured as mg GAE/mL necessary to inhibit the 50% of superoxide radicals. Quercetin was used as an antioxidant reference compound. Hydroxyl radical scavenging assay The HO scavenging capacity was measured as described by Cardozo et al. (2011). Antioxidant activity was measured using 2 mL cuvettes. The PEE (before and after simulated GD digestion) (0.5-10 mg GAE/mL) were added to a reaction mixture contained in KH 2 PO 4 /KOH buffer (pH 7.4), 50 mL of 10.4 mM 2-deoxy-D-ribose, 50 mL of 50 mM FeCl 3 , 50 mL of 52 mM EDTA. To start the Fenton reaction, 50 μL of 10 mM H 2 O 2 and 50 μL of 1.0 mM ascorbic acid were added and the reaction mixture was incubated 1 h at 37 C. Then, 500 μL of 2-thiobarbituric acid (1%, w/v) dissolved in trichloroacetic acid (3%, w/v) was added to the mixture and then, was incubated 20 min at 100 C. The reaction absorbance was read at 532 nm. The antioxidant activity was expressed as SC 50 values (mg GAE/mL necessary to inhibit by 50% the degradation of 2-deoxy-D-ribose by the HO ). Nitric oxide scavenging assay The ability of the chilto extracts to scavenging nitric oxide was carried out according to Govindarajan et al. (2003). Different concentrations of PEE (before and after simulated GD digestion) (10-60 μg GAE/mL), sodium nitroprusside (100 mM) and sodium phosphate buffer (0.2 M; pH 7.4) were incubated during 60 min at 37 C. Then, the Griess reagent was added and incubated for 15 min in the dark. The antioxidant capacity was measured spectrophotometrically at 550 nm. SC 50 was defined as the phenolic concentration (mg GAE/mL) necessary to scavenge 50% of nitric oxide. Ascorbic acid was used as a positive control. 2.12. Toxicity assays 2.12.1. Acute toxicity The acute toxic effect of the extracts was tested by the Artemia salina assay. After hatching, nauplius larvae with 24 h of incubation at 25 C were used for the assay. The larvae (10 per well) were transferred into wells of a microplate containing 100 μL of seawater. Then, increasing concentrations of extracts were added (2.5-250 μg GAE/mL). Negative control with DMSO and positive control with potassium dichromate (10-40 μg/mL) were assayed. After 24 h of exposition, the number of dead shrimps in each well was counted. Salmonella mutagenicity assay The mutagenic effect of red chilto skin, pulp and seed extracts were assayed on Salmonella Typhimurium strains TA98 and TA100 (Maron and Ames, 1983). Different concentrations of extracts (125-500 μg GAE/plate) were used. His þ revertant colonies were counted and compared to the revertant colonies of the controls (100 μL of DMSO). The reagent 4-nitro-o-phenylenediamine (4-NPD, 10 μg/plate) was used as a positive control. Extracts are considered mutagenic when the average number of revertants colonies was double or higher than the revertants colonies of controls. Three plates per experiment were assayed and two separate sets of experiments were performed in all cases. Statistical analyses Analyses were conducted at least three times with three different samples. Each experimental value is expressed as the mean AE standard deviation (SD). The one-way ANOVA with Tukey posttest at a confidence level of 95% was used for the comparisons between groups. For correlation analyses between total phenolic content and bioactivities Pearson's correlation coefficient (r) at p 0.05 was used. The analyses were developed using the statistic software InfoStat (Student Version, 2011). Results and discussion Chilto (orange and red) is a neglected and underutilized native species (NUS) in Argentine. The successful build-up as a commercially important crop requires the selection of cultivars with good fruit quality and the study of its potencialities as functional ingredient or food. For this reason, the study of red variety chilto grown in Argentinean Yungas and its comparison with the orange fruit is proposed. Parameters of physical-chemical quality of fresh fruit The quality of the fruits is the combination of characteristics, attributes and properties that give value to the food. The quality can also be defined as the set of parameters, such as soluble solids content (SSC), color and total acitdity that determine that a certain product is the taste of a consumer or a group that you want to satisfy. The CIELAB parameters of colour were taken into account to ensure the homogeneity and quality of the sample ( (Orqueda et al., 2017), showing differences in the shades of the epicarp among both varieties. The SSC value obtained for red fruits (9.20% AE 0.2) was lower than those for red tree tomatoes fruits from Ecuador and Spain (11.00-12.00%) (Vasco et al., 2009;Acosta Quezada et al., 2015). The red fruit of S. betaceum from Argentina showed lower total acidity (0.9%) than the orange fruit from the same place (1.9%) (Orqueda et al., 2017). The difference in the acidity of the two varieties is clear in the taste, so the red chilto have a more pleasant and less acid taste than the orange fruit. Metabolic profile of fresh red chilto In fruits, fructose, glucose and sucrose are the main soluble sugars and differ in the level of sweetness. The sugars are important in the flavor of the fruit causing a strong impact on the organoleptic quality of the fruits (D'Angelo et al., 2018). In fresh fruits of red chilto the level of fructose was 9.91 mg/g FW, glucose 7.32 mg/g FW and sucrose 6.62 mg/g FW (Table 1) and accordingly, the sweetness index of red fruits was 28.97. The sweetness index of chilto was higher than pear, apple and blueberries juice index (Rizzolo and Cortellino, 2018). Citric acid was the more abundant organic acid in the ripe red chilto (26.47 mg/g FW) followed by malic acid (4.86 mg/g FW). The level of GABA, glutamic and succinic acids were low, indicating the absence of GABA shunt pathway activity, and consequently, the lack of feeding to the tricarboxylic acid cycle, which diminishes the supply of NADH to the mitochondrial electron transport chain. Few free amino acids were detected, including L-asparagine and L-aspartic acid, allowing nitrogen assimilation and suppling oxaloacetate for the tricarboxylic acid cycle functioning, respectively. The level of ethanol was low indicating the absence of a fermentative process (Table 1). In red chilto fruit, the methanol content was 1 mg/g fruit, similar to that reported for apple (Possner et al., 2014). Methanol is present in pectin, a component of the plant cell walls and the middle lamella. Water-soluble pectin and methanol are released during processing of fruits and vegetables by enzymatic activities. To accurately measure the methanol contents of whole or processing fruits, it is important that pectin methyl esterase (PME) activity in the fruits is inactivated before analysis. In this way, when homogenizing the material, a rapid deesterification of the pectin will occur altering the amounts of methyl esters and free methanol. Chemical and functional characterization of chilto red fruit powder Removal of skin and seeds of fruits is a common practice by the Argentine food industry, and the population consumes only the fruits pulp (crude or cooked). For this reason, we analyzed the chilto skin, pulp and seed in order to give an added value to the waste material (skin and seed) in the industrial processing to obtain pulp, juice and other products. 3.2.1. Nutritional composition of the skin, pulp and seed powder 3.2.1.1. Macronutrients analysis. The soluble sugars content (8.04-20.5 g/100g DW) in the three fruit parts of the Argentinean red chilto was higher than in the orange chilto variety (2.23-11.91 g/100g DW) from the same ecoregion of Argentina (Orqueda et al., 2017). The carbohydrates content of red chilto fruits was lower than the ones reported for wild S. cajanumense of Ecuador (28.6 g/100 g DW) (Acosta-Quezada et al., 2015). The levels of glucose and fructose were higher in skin powder (1.02-3.23 g/100g) than in pulp or seeds powder (0.8-0.95 g/100g). These results were similar to those obtained for the orange chilto variety (0.90-2.72 g/100g) and much lower than for red tree tomatoes from Ecuador (9.0 g/100g) (Acosta-Quezada et al., 2015). The sucrose levels in ripe red chilto (4.23, 7.23 and 8.52 g/100g DW for skin, pulp and seeds, respectively) were higher than glucose (0.6-0.99 g/100g DW) and fructose (1.23-3.96 g/100g DW) ( Table 2). According to these results, the sweetness index of the three fruit parts was 6.77, 10.71 and 15.99 for skin, pulp and seeds, respectively, higher than in the orange chilto from the Argentine northwestern Yungas. Comparing with other flours or powders, the sweetness index of red chilto powder was lower than apple flours (59.85) (Pires et al., 2018). Digestible starch (DS) (0.02%, 0.01% and 0.006% for seeds, pulp and skin, respectively) and resistant starch (RS) (0.003% only in the pulp) were lower than for wheat flour (65% of total starch) (Swieca et al., 2017). In order to maintain normal postprandial glucose and fatty acid levels and avoid oxidative stress, a decrease in DS consumption must be achieved (Agama-Acevedo et al., 2018). In the three fruit parts, the protein level was between 5.2-16.3 g/100 g powder, higher than the content found in New Zealand fruits, whose values were between 1.9-2.0 g/100g (Athar et al., 2003). Although the protein content of the chilto powder was lower compared to soybean flour (43.2 g/100g) (Van Ee, 2009), it is relatively high for an unconventional flour. This data together with the low fat content (0.42-0.9 g/100 g), makes these unconventional flours or powders an ingredient of interest in the food industry. The total fiber level of skin, seed and pulp was high and similar to previously found in orange chilto (8.84-20.95 g/100g) (Orqueda et al., 2017). The dietary soluble fiber (DSF) content in red chilto skin, pulp and seeds was 9.72; 9.06 and 5.59%, respectively. The DSF content was higher than the reported in the flours of amaranth and quinoa (Kurek et al., 2018) and similar to wheat flour (Zhang et al., 2018). It is proven that dietary fibers produce certain healthy effects for consumers through increased fecal volume, stimulation of the fermentation of the colon, reduction of postprandial blood glucose and pre-prandial cholesterol level. These physiological effects are also well associated with reduced risk of cardiovascular diseases, cancer, diabetes, respiratory disease, infections, and others (Zhang et al., 2018). In conclusion, this study proved that chilto seeds and skin, two sub-products, as well as the pulp, could be exploited for their content in nutritional and functional components beneficial for human health. 3.2.1.2. Mineral analysis. The three chilto powders were high in potassium, with a content similar to banana (395 mg/100 g FW) (Livsmedelsverket, 2009) and low in sodium (below detection limits in pulp and seed powder) ( Table 2), even below the sodium content of the orange variety of chilto (Orqueda et al., 2017). This potassium/sodium ratio in red chilto powders is desirable in a healthy diet. The magnesium and calcium contents were higher than in tree tomatoes from Ecuador (Vasco et al., 2009), principally in seed powder. Compared with orange chilto, the magnesium, potassium and calcium content was lower in red chilto, although the iron content was higher in the seeds of red chilto (Orqueda et al., 2017). Since iron is an important mineral for the prevention of anemia due to iron deficiency, and the diverse clinical manifestations associated with this pathology, the content of this microelement converts chilto powder as an interesting functional ingredient (Abbaspour et al., 2014). However, other studies on the bioavailability of this mineral in chilto powder would be necessary. Phytochemical composition of the skin, pulp and seed powder 3.2.2.1. Polyphenolic compounds content. Plant foods have large amounts of phytochemicals. There is evidence on the benefit that the consumption of foods containing phenolic compounds is associated with the prevention of chronic disorders such as cardiovascular diseases and type 2 diabetes (Costamagna et al., 2016). In recent times, the interest in these bioactive compounds has increased, as well as their impact on health. The content of phenolic compounds in red chilto was 408.9 AE 2.3, 334.0 AE 1.2 and 623.6 AE 1.6 mg GAE/100 g DW in the skin, pulp and seeds, respectively. The phenolic profile of pulp and skin was very similar. Two main compounds, rosmarinic acid and caffeoylquinic acid, were identified ( Figure 2). These two hydroxycinnamoyl derivatives had been reported previously in the orange chilto from Argentina Yungas and in yellow and purple varieties from Ecuador (Espin et al., 2016;Orqueda et al., 2017) but in orange chilto the diversity of phenolic compounds was greater (12 caffeic acid derivatives and related phenolics, 10 rosmarinic acid derivatives and 7 flavonoids), (Orqueda et al., 2017). The total content of each phenolic acid (rosmarinic acid and caffeoylquinic acid) in red chilto skin powder was 0.291 AE 0.001 g/100 g powder and 0.354 AE 0.002 g/100 g powder, respectively. Several biological activities are attributed to these phenolic acids, such as antioxidant, anti-inflammatory, immunomodulatory, hypoglycemic, hypolipidemic, antimutagenic, antitumor, antimicrobial, hepatoprotective, neuroprotective, and nephroprotective properties (Spínola and Castilho, 2017;Costa et al., 2014). The food preferences of the population are characteristic of each region and can define the intake of polyphenols in each of them. However, although the recommended intake depends on the region, preferences, age, among others, the presence of these bioactive compounds enhances the nutritional value of the fruits. Regarding diet contribution of phenolics, 100 g of fresh red chilto contributes 84.87 mg of phenolic compounds, including 43.96, 29.79 and 11.12 mg of these compounds in the seeds, pulp and skin, respectively. Hydrolyzable tannins were not detected in any of the three powders. But, all red chilto powders contain condensed tannins (Table 3), mainly in the seed powder (265.0 mg/100 g DW). This is the first report on the content of condensed tannins for chilto. Condensed tannins are biofunctional compounds of interest for their multiple health attributes that include anti-inflammatory, antimicrobial, antiviral, antitumoral, hypocholesterolemic, and digestive enzymes (α-amylase, α-glucosidase and lipase) inhibitory activities and thus, their benefits in cardiovascular diseases, diabetes, and gastrointestinal disorders . Anthocyanins are a group of flavonoid compounds responsible of the colours of many plant structures (flowers and fruits) and are related with health-related properties such as effect reducing the risk of cardiovascular diseases (De Pascual-Teresa and Sanchez-Ballesta, 2008). In red chilto, the content of anthocyanins was 10.6, 62.5 and 65.23 mg EC/100 g DW for the pulp, skin and seeds, respectively. The total content of pigment (60.38 mg EC/100 g FW) in red chilto was much higher than the reported to orange chilto (1.78 mg/100 g, only in the skin) from Argentinean Yungas (Orqueda et al., 2017) and similar to that reported in blackberry (58.61 mg/100 g FW; Souza et al., 2014). These results provide evidence of difference in profile and quantities of phenolic compounds among the orange and red varieties. This information is essential for the development of selection and breeding programmes. 3.2.2.2. Vitamins. The ascorbic acid (vitamin C) content of the skin, pulp and seed powders is shown in Table 3. The results show that the powders are rich in vitamin C, a potent antioxidant compound with large biological activities. The vitamin C content of the red chilto powder was similar to orange chilto from Argentina and was higher than that of the edible tissues of oranges Citrus sinensis (Abeysinghe et al., 2007). Carotenoids were only detected in red chilto pulp (0.9 mg/100 g DW) while en orange chilto from Argentina was found in the three parts. These pigments have multiple biological properties such as oxidative stress protection, anticarcinogenic capacity, macular degeneration prevention or bone health support (Hornero-Mendez et al., 2018). Functional properties of seed, skin and pulp powders and PEE The powders of pulp, skin and seed of chilto were submitted to simulated GD. Due to the extensive evidence that phenolic compounds are effective antioxidants and inhibitors of digestive enzymes, and the intention to also use the fruit powders as raw material for phenolic extracts preparation, the effect of gastroduodenal digestion was studied also on phenolics enriched extracts. 3.2.3.1. Hypoglycemic effect of chilto powder and PEE. Metabolic syndrome is a combination of at least three of the cardiovascular risk factors: obesity, dyslipidemia, hypertension and hyperglycaemia or Type 2 diabetes. All these risk factors are related to insulin resistance, oxidative stress and an inflammatory state. The level of postprandial hyperglycemia can be controlled by inhibiting the activities of α-amylase and α-glucosidase, slowing the diffusion and adsorption of glucose in the intestinal epithelium, stimulating insulin secretion by pancreatic cells and inhibiting the transporters of glucose at the cellular level (Costamagna et al., 2016). The commercial medicines used for treating diabetes may cause several side effects, including flatulence and diarrhea. Therefore, it is necessary to investigate natural sources of food for the prevention of postprandial hyperglycemia. 3.2.3.1.1. Effect of chilto powder on glucose adsorption, diffusion and transport. Adsorption The glucose adsorption capacity of red chilto powders before and after GD was assayed. The seed powder showed higher adsorption capacity than skin and pulp powders (Figure 3). No significant differences were observed between the glucose adsorption capacity of digested and undigested seed and pulp powders. The adsorption capacity of the digested chilto skin powder was greater than undigested. The results revealed that the adsorption capacity of chilto seeds powder was higher than acarbose, wheat bran and oats at a concentration of 10 mM of glucose (Ahmed et al., 2011) and similar to Caesalpinia bonducella and Myristica fragrans seed (Bhutkar et al., 2018). These results indicated that the red chilto powders could decrease the amount of glucose available to be absorbed by the intestinal epithelium resulting in normal blood glucose values. Furthermore, condensed tannins, which are also present in red chilto powders, principally in seed powders could bind to the enzymes at non-specific sites, diminishing their activities and have been proposed as non-competitive inhibitors of these digestive enzymes (Barrett et al., 2018). Diffusion The effect of chilto powder before and after GD on retarding glucose diffusion across the dialysis membrane is represented in Figure 4. During the study, the movement of glucose across the dialysis membrane was monitored every 30 min for 240 min, and it was observed that the powder samples delayed the movement of glucose to the dialysate through the membrane compared to the control. The rate of glucose diffusion across the dialysis membrane was increased over time, but all the samples were able to inhibit glucose diffusion compared to the control (Figure 4). No significant differences were observed between the digested and undigested powders. These effects could be attributed to the network formed by the fibers where glucose is trapped (L opez et al., 1996). The dietary fiber of the powders forms viscous gels in contact with water, which slows the access of glucose to the small intestine epithelium, decreasing its absorption. 3.2.3.1.2. Effect of chilto PEE on glucose uptake by S. cerevisiae cells. The yeast model to evaluate the mechanism of glucose transport through the cell membrane is a test for the in vitro evaluation of the possible hypoglycemic effect. The results of the rate of glucose transport across the cell membrane in yeast cell system are represented in Figure 5 I. The presence of phytochemicals of chilto increase the uptake of glucose by the yeast cells, therefore the amount of glucose remaining in the medium is lowest after the incubation time. The area under the curves of glucose uptake obtained with polyphenols from seed powder (5901.52 AE 124.79) was higher than pulp (3504.01 AE 441.78), followed by skin (2789.50 AE 109.58). The chilto seeds polyphenolic extract was more effective to increase glucose transport up to 81% compared to the control ( Figure 5I). Enhancement on glucose uptake by low concentration of seeds polyphenols was similar to chia and cinnamon extracts (Woldemariam and Van Winkle, 2015). Furthermore, SEM photographs showed that the cell surface structure of the yeasts was not affected in the presence of chilto extracts ( Figure 5II). These results show that the chilto polyphenols could favor the entry of glucose into the cells, resulting in a decrease in blood glucose levels. Further in vivo experiments should be achieved in order to confirm that these natural extracts are possible antidiabetic nutraceuticals. 3.2.3.1.3. Inhibition of carbohydrate metabolism enzymes by PEE. Pancreatic α-amylase and α-glucosidase hydrolyse complex carbohydrates and disaccharides in absorbable monosaccharides in the intestinal epithelium (Costamagna et al., 2016). An objective in the treatment of type 2 diabetes focuses on the inhibition of these enzymes as a way to lower blood glucose levels (Zhang et al., 2015). Anti-diabetic agents that are used in clinical practice, such as acarbose, voglibose, and miglitol are competitive inhibitors of these enzymes which delay the hydrolysis of carbohydrates and thus alleviate postprandial hyperglycemia (Scheen et al., 2015). However, the use of these drugs has some disadvantages including drug resistance, side effects, and even toxicity (Spínola et al., 2020). Due to their strong inhibitory activities, these enzyme inhibitors can produce several side effects, since dangerous ones like the excessive lowering of blood sugar levels when used in combination with other medications used to treat diabetes, to banaler such as abdominal pain, diarrhoea, and flatulence. In this sense, new natural inhibitors with moderate activities and good bioavailability can be useful to manage obesity, insulin resistance or type-2 diabetes. Bioavailability is dependent upon their release from the plant matrix (bioaccessibility), their stability during digestion process, and the efficiency of their transepithelial passage (Manach et al., 2004). So, the inhibitory capacity of amylase and glucosidase of red chilto polyphenolic extracts was determined before and after in vitro gastroduodenal digestion to determine the stability during digestion. All extracts enriched in polyphenols of red chilto were active to inhibit both enzymes with IC 50 values between 25.0-89.3 μg of GAE/mL for α-glucosidase and 50.0-136.5 μg of GAE/mL for α-amylase (Table 4). A strong correlation (r 2 ! 0.84) between total phenolic contents and the inhibitory capacity of the enzymes (α-glucosidase and α-amylase) was observed with seeds, skin, and pulp extracts (Table 5). However, the correlations were only significant (p < 0.05) with seeds and pulp extracts against α-glucosidase. A very strong significant correlation between total ; seed ; pulp ; and skin . The symbol (*) indicates significant differences between seed and pulp extracts compared to skin extracts at a given time. phenolic concentrations and α-glucosidase inhibition was observed with these two extracts (r 2 ! 0.96, p < 0.05). The seeds powder was the most active extract, probably due to the higher contents of total phenolics and condensed tannins, the last ones recently proved to be effective inhibitors of α-glucosidase and α-amylase (Barrett et al., 2018). The PEE of red chilto were more active towards the α-glucosidase than the acarbose reference compound but had a lower effect on α-amylase. Also, it is noteworthy that the extracts obtained from the orange variety flours proved to be much more active than those of the red variety, probably due to its less phenolic components diversity than orange chilto (Orqueda et al., 2017). Besides, there were no significant differences after subjecting the extracts to the digestive process, contrariwise to the PEE of seeds of orange variety (Table 4). According to the literature, CQA and RA which are present in red chilto have a great capacity for inhibition of digestive enzymes (Lin et al., 2011;Spínola and Castilho, 2017). Previous reports showed that the concentration of rosmarinic acid is slightly reduced after gastric digestion and that was more stable in the basic pH of the pancreatic environment than in the acidic pH of the gastric juice (Costa et al., 2014;Porfirio et al., 2010). On the other hand, isomerization and hydrolysis of CQA and partial opening of lactones intramolecular ring were observed after incubation of CQA with artificial and human intestinal fluids (pH from 6.3 to 7.7), with slightly losses after 8-h incubation (Farah and Duarte, 2015). The results would indicate that the inhibitory effect could be attributed to both phenolic acids and other minoritary compounds, non-detectable with the used methodologies. Effect of chilto PEE on in vitro fat absorption. Obesity has become a serious health issue around the world due to the connection with a large number of pathological disorders, including metabolic syndrome, diabetes mellitus, hypertension, atherosclerosis and cancer. Inhibition of dietary fat absorption is one of the common approaches to decrease excessive energy intake. Pancreatic lipase hydrolyses lipids in absorbable fatty acids and monoglycerides and, therefore, their inhibition is a key target in the prevention and treatment of obesity. Orlistat is a commonly used antiobesity drug and the reference compound for the inhibition of pancreatic lipase, but it produces several adverse effects. Again, the potency of available commercial inhibitors is mostly responsible for these undesirable effects. Hence the importance of finding moderate lipase inhibitors and natural inhibitors from plant arise as an alternative. In this regard, the inhibitory activity of red chilto PEE was assessed on pancreatic lipase (Table 4) before and after GD digestion. The highest inhibitory activities were observed in the PEE from seeds and pulp before GD digestion (IC 50 : 4.5 and 4.7 μg GAE/mL, respectively). These both extracts showed a very strong significant correlation between total phenolic contents and the ability to inhibit pancreatic lipase (r 2 ! 0.97; p < 0.05) ( Table 5). The inhibitory capacity of red chilto ethanolic extracts was similar to that of extracts of orange chilto and Geoffroea decorticans (IC 50 : 4.0 μg GAE/mL) (Costamagna et al., 2016;Orqueda et al., 2017). The ability to inhibit lipase exhibited by the PEE from red chilto could be related to the presence of functional compounds such as caffeoylquinic Table 4. Effect of phenolic-enriched extract (PEE) of Argentinean Solanum betaceum red fruit powder (seed, pulp and skin) before and after simulated gastroduodenal digestion (GD) on enzymes related to carbohydrate and fat metabolism and oxidative processes. Reference compounds are included for comparison. S. betaceum Enzyme acids or rosmarinic acid which were previously described as pancreatic lipase inhibitors (Spínola and Castilho, 2017). However, after gastroduodenal digestion, the inhibitory activity of the red chilto PEE decreases by more than 20%, Table 4. The decrease in inhibitory activity after GD digestion may be related to changes in the identity and quantity of bioavailable polyphenols during the passage through the gastrointestinal tract (Costa et al., 2014;Porfirio et al., 2010;Farah and Duarte, 2015). 3.2.3.3. Antioxidant activity of chilto PEE. Several studies indicate that certain complications associated with the metabolic syndrome such as atherosclerosis, hypertension, visceral adiposity and insulin resistance may be associated to an increase in oxidative stress (Costamagna et al., 2016). Some researchers suggest that oxidative stress could be an early event in the pathology of these chronic diseases rather than merely a consequence or an innocent bystander. Considering this, the antioxidant activity of the three PEE was assessed before and after simulated GD (Table 4). Most of the extracts exhibited very strong significant correlation between total phenolic contents and antioxidant activities (r 2 ! 0.95; p < 0.05) ( Table 5). The extracts were able to scavenge ABTS radical with IC 50 values between 3.4-8.2 μg GAE/mL. Also, the PEE were very efficient protectors of RBC membrane (0.7-3.6 μg GAE/mL) with major activity of pulps extracts (1 and 0.7 μg GAE/mL, before and after GD digestion). The activity was much higher than extracts of other fruits known for its capacity for the prevention of oxidation and cellular damage, such as quince (Cydonia oblonga) or sweet cherries , H 2 O 2 and NO. However, the HO radical is considered the most harmful ROS due to its extreme reactivity and oxidative potential, even reacting with compounds such as alkanes (Cardozo et al., 2011). The highest inhibitory potency recorded against this radical was observed in skin powder and was consistent with its higher contents of flavones and flavanols (Table 4). Simulated GD did not significantly affect the protective activity of biomolecules and ROS/RNS scavenging capacity of the PEE. Again, rosmarinic acid and caffeoylquinic acids could be contributing to the functional property of the PEE. Antioxidant capacities of these phenolic compounds were repeatedly demonstrated in both in vitro and in vivo experiments (Zhang et al., 2015;Markovi c and To sovic, 2016;Adomako-Bonsu et al., 2017). Acute toxicity and genotoxicity tests Toxicity tests are defined as the qualitative and quantitative studies of the deleterious effects caused by chemical or physical agents on the structure and function of living systems. These assays are used for the evaluation of safety and prevention of damage to man. In this sense, the toxicity of S. betaceum extracts was evaluated. None of the extracts of the three parts of the fruit produced significant changes in the number of S. Typhimurium colonies compared to the control at the concentration of 500 μg GAE/plate (data not shown). This result indicates that none of the extracts has genotoxic effects in the cellular systems used, suggesting the absence of compounds that can cause mutations of the frameshift type (TA98) or substitution of bases (TA100). Also, the polyphenols enriched extracts of red chilto had not toxic effect on Artemia salina at the concentrations tested with respect to the negative control (DMSO). With the highest concentration tested (1000 μg GAE/mL) the viability of A. salina was not affected. Besides, no changes in larval movements were observed after the incubation period. Conclusions The red chilto fruits are a source of appreciated bioactive compounds, such as the rosmarinic acid, caffeoylquinic acids, condensed tannins, anthocyanins, ascorbic acid and soluble dietary fiber. These valuable compounds are also present in the skin, pulp and seed powders, turning them into an excellent source of nutraceuticals and, thus, into dietary supplements or functional foods useful for the management of metabolic syndrome and its related pathologies. Red chilto fruits can alternatively be used as flours or powders to prepare diverse foods or as the raw material for the preparation of functional phenolic extracts. The (before and after GD digestion) seed powder showed high glucose adsorption capacity. The polyphenolic extracts obtained from seed flours were the most active for most of the antioxidant tests as well as to inhibit the enzymes related to the development of the metabolic syndrome and could favor the entry of glucose into the cells, resulting in a decrease in blood glucose levels. The high content of phenolic compounds with high antioxidant activity in seed extracts make them a good functional food candidate and could serve as an ingredient for the preparation of nutraceuticals for human health improvement. Skin powder was also active as an antioxidant supplement. These results provide added value to the different fruit parts that are normally discarded during industrial processes. Pulp powder is attractive because of the presence of bioactive compounds such as phenolic compounds, carotenoids and the high amount of vitamin C, which makes it an excellent dietary supplement with nutritional and phytochemical qualities. The results of the present work showed that both red chilto and the previously studied orange variety are potent inhibitors of digestive enzymes related to metabolic syndrome. The polyphenols of the orange variety displayed higher α-amylase and α-glucosidase inhibitory capacity, while the PEE of the red fruits showed higher inhibition of pancreatic lipase. These results highlight the use of both powder and polyphenolic extracts of red chilto in metabolic syndrome. Besides, natural products often have the advantage of having moderate activity, a wide spectrum of possible multiple/synergistic effects and have fewer side effects with low toxicity, which makes them of special interest for the development of alternative "biologically active products". However, further research including in vivo studies and clinical trials are necessary to confirm the potential healthy-beneficial effects of red chilto fruits and its powders. Author contribution statement Maria E. Orqueda: Performed the experiments; Analyzed and interpreted the data; Wrote the paper. Sebastian Torres, Iris C. Zampini, Estela M. Valle, Guillermo Schmeda-Hirschman, Maria I. Isla: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper. Florencia Cattaneo: Performed the experiments; Analyzed and interpreted the data.	2020-02-13T09:22:24.547Z	2020-02-01T00:00:00.000	{ "year": 2020, "sha1": "f274ae3ad6b3ea48862782ad477cc1e8324c9c83", "oa_license": "CCBYNCND", "oa_url": "http://www.cell.com/article/S2405844020302322/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "58d8f4cbc910ae3301f821de260b7c4027dbd8c1", "s2fieldsofstudy": [ "Agricultural And Food Sciences" ], "extfieldsofstudy": [ "Medicine", "Chemistry" ] }
233647255	pes2o/s2orc	v3-fos-license	Endophthalmitis panorama in the Jerusalem area The spectrum of microbial infections and the pattern of their susceptibility are variable among communities. Researching these data will lead to the establishment of the most appropriate national management strategies. The purpose of this study was to analyze the epidemiological, clinical, microbial spectrum and antibiotic susceptibility of endophthalmitis cases in a tertiary referral center in Jerusalem. Retrospective review of medical charts of patients presenting with endophthalmitis over a 12-year period. A total of 74 eyes of 70 patients (males 56%) were included. Mean age ± SD at presentation was 60 ± 19.5 years. Exogenous endophthalmitis accounted for 78% of cases, of which 62% followed an intraocular surgery, 21% occurred after intravitreal injections, 10% followed infectious keratitis and 7% were posttraumatic. Endogenous cases were predominantly observed in diabetic patients. Microbial isolates were identified in 44 samples. Of them, gram-positive bacteria were the predominant microorganisms detected in 33 samples (75%); Staphylococcus epidermidis and Enterococcus faecalis were the most commonly detected pathogens. Mean presenting ± SD LogMAR visual acuity (VA) was 2.38 ± 1.21 and it improved at last follow-up to 1.7 ± 1.37 (p = 0.004, paired t test). Cases secondary to gram-positive microbes were associated with improved VA during the follow-up while cases secondary to gram-negative microbes was correlated with poor final VA (p = 0.046, r2 = 0.4). There was no evidence of bacterial resistance in the antibiograms for either vancomycin, ceftazidime, ceftriaxone or amikacin. Intraocular surgery remains the most common event preceding endophthalmitis with coagulase-negative staphylococci being the most frequently detected microorganisms. The microbial spectrum of endophthalmitis is similar to that in the western world. Introduction Endophthalmitis is one of the most dreadful entities in Ophthalmology. Whether of exogenous or endogenous origin, the consequences for the visual function and globe integrity could be poor. The infection involves the inner layers of the eye producing a vitreous exudate that characterizes the disease. The majority of cases are related to intraocular surgery. Exogenous cases especially those related to trauma and endogenous cases pose a challenge for the ophthalmologist who needs to institute prompt and effective treatment in order to obtain the best possible results [1][2][3]. Characterizing the epidemiological data is important in order to reveal the spectrum of the most common pathogens, their susceptibility/resistance profile, the predisposing conditions and to determine the best empirical antibiotic regimen to use. These data will lead to the establishment of the most appropriate management strategies. The spectrum of microbial infections and the pattern of their susceptibility are variable among communities [1][2][3]. To date, there are no publications that studied the microbial spectra and the antibiotic susceptibility in the Jerusalem area. The aim of the present study is to analyze the epidemiological, clinical, microbial spectra and antibiotic susceptibility of patients diagnosed with endophthalmitis in a tertiary referral center in Jerusalem. Methods This is a retrospective study in which the medical records of patients diagnosed with endophthalmitis between the years 2007-2019 were reviewed. Data retrieved from the medical files included demographic features, year of presentation, type of endophthalmitis (exogenous, endogenous), the inciting event in exogenous cases and the underlying illnesses in exogenous and endogenous endophthalmitis, systemic and local therapy, surgical management including lapse of time between presentation and surgical intervention, specific microorganism isolated by cultures and susceptibility according to the antibiogram. The ocular examination included collection of vitreous and/or aqueous humor samples either by pars plana vitrectomy or tap. The samples were inoculated onto blood agar, chocolate agar and thioglycolate broth and incubated at 37°C for bacterial isolation. Until 2012, bacterial isolates were identified by classical biochemical methods (sugar fermentation tests). From 2012 and on, isolates were identified mainly by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDi TOF-MS, VITEK MS, bioMérieux, France). Antibiotic susceptibility testing of the isolates was performed using disk diffusion method, E-test or VITEK MS. The antibiotic susceptibilities were determined according to the criteria of the Clinical and Laboratory Standards Institute (CLSI). Samples were concomitantly evaluated for fungal infections. They were stained by calcofluor white dye and were cultured on sabouraud dextrose agar and incubated at 25°C for up to 2 weeks. From 2007 to 2012, identification of Candida and filamentous fungi was based on microscopic appearance and the use of CHROMagar culture medium and API ID 32C (bioMérieux, France). From 2012 and on, isolates were identified mainly by Matrix-Assisted Laser Desorption Ionization Time-of-Flight mass spectrometry (MALDi TOF-MS, VITEK MS, bioMérieux, France). Susceptibility testing was performed using the E-test method according to manufacturer's instructions (bioMérieux, France). Presenting visual acuity (VA) and VA at one month and at last follow-up, ophthalmic manifestations and ocular complications were recorded. LogMAR (log of the minimum angle of resolution) notation was used to compute the change in VA. VA of B 6/60 was defined as severe visual loss, VA of [ 6/60-\ 6/12 was defined as moderate visual loss and VA of C 6/12 was defined as good VA. The institutional review board of the hospital approved the study, including waiver of informed consent for this chart review study. The study was conducted adhering to the tenets of the Declaration of Helsinki. Statistical analysis Statistical analyses were performed using SPSS (version 26.0; IBM Corp., Armonk, NY). Measure of VA was summarized using means, standard deviation (SD) and range. Test for normality of data was first performed using the Kolmogorov-Smirnov and Shapiro-Wilk analysis. As the data were normally distributed, the comparison of VA at baseline and during the follow-up was performed using paired student's t test. Where appropriate, sensitivity analysis using nonparametric Wilcoxon-signed rank tests or chi-square was performed. A univariate regression analysis was used to assess the correlation between the final VA and the microbe type, positivity of blood and vitreous cultures, presenting signs and symptoms, source and etiology of the infection, immune system status and treatment modality. In case both variables appeared to be statistically significant, a multivariate regression analysis was carried out. p value \ 0.05 was considered as statistically significant. Demographic features A total of 74 eyes of 70 patients (39 males (56%)) diagnosed with endophthalmitis based on the clinical presentation were included. Mean ± SD age at presentation was 60 ± 19.5 years (range 3-90). Endophthalmitis was of an exogenous etiology in the majority of eyes (58 eyes, 78%). Mean follow-up time was 24 months (range 1 month-8 years). Positive vitreous cultures were obtained in 35 (60%) out of the 58 eyes (57 patients), all of which indicated bacterial infection. Eleven patients (19%) had diabetes mellitus type 2 (DM2) and two were on immunosuppressive medications because of lupus in one patient and after bone marrow transplantation in another patient. Positive vitreous cultures were obtained in nine eyes (56%), of them; fungal infection was detected in five eyes (56%) and bacterial infection in four eyes (44%). Fungal isolates included Candida in four samples and Pseudallescheria boydii in one sample (Tables 2 and 3). Bacterial infections included Streptococcus anginosus, Streptococcus salivarius, Staphylococcus aureus and Staphylococcus epidermidis. Three patients had bilateral EnE ( Table 2): inpatient #4, Candida albicans was isolated from the vitreous sample of one eye; in-patient #9, it was presumed to be fungal because of the compatible clinical findings on funduscopy; and in-patient #10, it was secondary to Methicillin-resistant staphylococcus aureus (MRSA) isolated from blood sample. Microbial spectrum Table 3 presents the microbial isolates from the 44 (59%) culture-positive samples. Gram-positive bacteria were the most commonly isolated microorganisms, detected in 33 samples (75%). Gram-negative bacteria were isolated in six (14%) and fungi were present in five samples (11%). Among all the 44 isolates, Staphylococcus epidermidis was the most commonly isolated microbe present in 14 samples (32%). Enterococcus faecalis was the second most commonly detected microbe observed in four samples (9%). Clinical presentation All patients were symptomatic. The majority of patients (41 patients, 59%) presented with a combination of symptoms. The most common presenting symptom was deterioration of vision reported by 50 patients (71%). Thirty-four patients (49%) presented with pain, 30 patients (43%) presented with eye redness and 14 patients (20%) presented with eye discharge. Mean presenting LogMAR ± SD VA was 2.38 ± 1.21, and even though it remained poor throughout the entire study, it improved at one month Univariate and multivariate analysis were used to assess the correlation between last VA and microbe type, positivity of blood and vitreous cultures, presenting signs and symptoms, source and etiology of the infection, immune system status and treatment modality. In those models, cases secondary to grampositive microbes were associated with improved VA during the follow-up while cases secondary to gramnegative microbes were correlated with poor final VA (p = 0.046, r 2 = 0.4). Figure 1 shows the changes in VA throughout the study period per microorganism type. Endophthalmitis following intravitreal injection versus following cataract operation Twelve patients (mean age at presentation 57.9 ± 15.6 years) presented at a mean of 5.1 ± 3.9 days after intravitreal injection (IVI). Mean presenting LogMAR ± SD VA was 2.3 ± 1.2 and it was 1.7 ± 1.4 and 1.3 ± 1.5 one month later and at last follow-up, respectively (p = 0.14 between baseline and last follow-up). The most common microbe was Staphylococcus epidermidis isolated in 10 eyes. Enterococcus faecalis was isolated in one eye, and in one eye, no microbe was isolated. Seventeen patients (mean age at presentation 69.3 ± 12.5 years) presented at a mean of 34.5 ± 60 days following phacoemulsification and IOL implantation. Mean presenting LogMAR ± SD VA was 2.7 ± 1.2, and it was 1.8 ± 1.7 and 1.8 ± 1.8 one month later and at last follow-up, respectively (p = 0.10 between baseline and last follow-up). Enterococcus faecalis was isolated in 3 eyes, Staphylococcus epidermidis in 2 eyes, and Streptococcus agalactiae, Streptococcus mitis, Granulicatella adiacens, Mycobacterium fortuitum, Proteus mirabilis and Pseudomonas aeruginosa in one eye each. In 6 eyes, no pathogen was isolated. There was no difference in presenting and last VA between eyes that developed endophthalmitis post-IVI and eyes that developed endophthalmitis after cataract operation. Delayed-onset endophthalmitis Fifteen patients (15 eyes, 20.3% of all eyes), with a mean age at presentation of 62.3 ± 17.4 years, presented with delayed-onset endophthalmitis (DOE) (C 6 weeks after the preceding surgical intervention). Eight eyes (53%) had a preceding glaucoma interventional procedure, 6 eyes (40%) presented after phacoemulsification, IOL implantation with or without vitrectomy and one eye presented after penetrating keratoplasty. Patients presented at a mean of 45.2 ± 63.8 months (range 2 months-18 years) following the inciting procedure. Patients who developed DOE following glaucoma interventional procedures presented significantly later than patients who developed DOE following a non-glaucoma interventional procedure (81 ± 70.8 vs. 4.3 ± 1.8 months, respectively, p = 0.013). Mean presenting LogMAR ± SD VA for all eyes with DOE was 2.4 ± 1.4, and it significantly improved after one month and at last follow-up to 1.2 ± 1.2 and 1.2 ± 1.2 respectively, (p = 0.01 from baseline to last follow-up). Positive vitreous cultures were obtained in 6 of the eyes (40%), with Mycobacteria and Streptococci cultured in 2 eyes each and Moraxella nonliquefaciens and Staphylococcus epidermidis in one eye each. Eyes that developed DOE following a preceding glaucoma procedure had a mean presenting LogMAR VA of 2.8 ± 1.3 and it showed a trend of improvement to 1.2 ± 1.2 and 1.6 ± 1.4 at one month and at last follow-up, respectively (p = 0.07 from baseline to last follow-up). Eyes that developed DOE following a non-glaucoma intervention had a mean presenting LogMAR VA of 1.8 ± 1.4 and it was 1.1 ± 1.3 and 0.7 ± 0.8 after one month and at last follow-up, respectively (p = 0.16 from baseline to last follow-up). Patients with early-onset endophthalmitis (within 6 weeks of the preceding surgical intervention), with a mean age of 61.6 ± 20 years at presentation, presented at a mean of 6.8 ± 6.1 days following the inciting procedure. Mean presenting LogMAR VA was 2.6 ± 1.2 and it was 1.9 ± 1.5 and 2.1 ± 1.6 at one month and at last follow-up, respectively, (p = 0.11 from baseline to last visit). There was no difference between early-onset and delayed-onset endophthalmitis with regard to presenting and last VA. Ocular complications The most common complication was rhegmatogenous retinal detachment, documented in 12 eyes (16%). Phthisis bulbi developed in 12 eyes (16%), five of which were eviscerated/enucleated ( Table 5). The mean age ± SD of patients who developed phthisis bulbi was 68 ± 20 years (females 58%). The preceding events were cataract operation in four eyes, infectious keratitis in three eyes, glaucoma filtering valve in two eyes, intravitreal injection in two eyes and endogenous fungal endophthalmitis after near-drowning in one eye. In six eyes (50%), no pathogen was isolated. The most commonly isolated microorganism was Enterococcus faecalis in two eyes. Mycobacterium chelonae, Streptococcus mitis, Streptococcus agalactiae, and Pseudallescheria boydii were each isolated in one eye. Of the five eyes that were enucleated/eviscerated, no growth was documented in three eyes whereas Mycobacterium chelonae and Pseudallescheria boydii were detected in the remaining two eyes. Amikacin monotherapy (0. 4 mg/0.1 ml) was injected to one eye (1.3%) based on a positive vitreous culture result of amikacin-sensitive Pseudomonas aeruginosa from another hospital. Voriconazole monotherapy was injected to one eye (1.3%) of a patient with candidemia. One eye (1.3%) did not receive intravitreal therapy given the late presentation and the lack of potential vision that necessitated evisceration ( Table 2, pt #13). Intravitreal dexamethasone (0.4 mg/0.1 mL) was administered to 22 eyes (31%) in combination with intravitreal antimicrobial therapy. The mean number of intravitreal antimicrobial injections was 1.77 (range 1-5). Systemic intravenous antibiotics were part of the treatment regimen in 44 patients (63%) (All patients with EnE received systemic antibiotics). Vancomycin was the most commonly used systemic antibiotic treatment, administered to 25 patients; Ceftazidime was used in 16 and ceftriaxone in 11 patients. Oral prednisone was administered in 13 patients (19%). Pars plana vitrectomy was performed in 57 eyes (77%) at a median of 1 day after presentation. In the remaining eyes, tap and inject was performed. Antibiograms did not reveal resistance to the used antimicrobial agents (Table 6). Discussion The present study is the first study in Israel that aimed to portray the microbial spectra of endophthalmitis cases and their susceptibilities to antimicrobial agents. It is also the first study that aimed to study the demographic, clinical and microbiological characteristics of patients presenting with endophthalmitis in the Jerusalem area. Jerusalem has currently a population of 927,000 inhabitants making it the biggest city in Israel (https://www.cbs.gov.il/he/mediarelease/ DocLib/2019/156/11_19_156b.pdf). The present study highlights the following points: First, ExE post-cataract operation was the leading cause of endophthalmitis. All ExE cases were caused by bacterial microorganisms mainly gram-positive bacteria, of which Staphylococcus epidermidis was the most common pathogen. Second, of all the isolates, gram-positive bacteria accounted for 75% of them whereas gram-negative bacteria and fungi accounted for 14% and 11% of them, respectively. Third, EnE was mainly caused by fungal infection in immunocompromized patients, of which candida was the predominant pathogen. Fourth, all gram-positive bacteria were susceptible to vancomycin. Fifth, treatment resulted in a significant drop in the percentage of eyes with poor VA and evisceration/enucleation were performed in five eyes only. Finally, there was no evidence of bacterial resistance in the antibiograms for Similar to our study, Gentile et al. reported [3] that gram-positive bacteria were the most common isolates followed by gram-negative bacteria in 10.3% and (Table 7). Schimel et al. [4] and Liu et al. [2] found that gram-positive bacteria were the predominantly isolated pathogens in 72.9% and 65%, respectively, however, fungi were the second most prevalent pathogens in 15.8% and 18.4%, respectively, and gram-negative bacteria ranked third in 10.7% and 16.6%, respectively. On the other hand, Lin et al. [6] and Satpathy et al. [5] reported a higher percentage of gram-negative bacteria in their studies, 27.8% and 38%, respectively. Lin et al. [6] reported that gram-negative bacteria accounted for 26% and 30% of the microbial isolates that caused posttraumatic and keratitis-related endophthalmitis, respectively. In the present study, infectious keratitis and trauma were the least common predisposing causes of exogenous endophthalmitis accounting only for 10% and 7% of the cases, respectively. This could explain the much lower isolation of gram-negative bacteria in our study. Cases secondary to gram-negative microbes were correlated with poor final VA, similar to the results of the endophthalmitis vitrectomy study group [7]. Fungal endophthalmitis was the least frequent microbial cause of endophthalmitis in the present study (11%), typically affecting immunocompromised patients. The most common infection was candida while aspergillus was not detected in the present study. Lin et al. [6] reported on fungal endophthalmitis as the cause of 24.3% of endophthalmitis cases. Filamentous fungi accounted for 57.6% of isolates causing keratitis-related endophthalmitis. Fusarium and aspergillus species caused 24.24% and 21.21%, respectively, of all cases of keratitis-related endophthalmitis. [6] The low rate of keratitis is the most likely explanation to the absence of these pathogens in our case series. In the present study, one patient sustained Pseudallescheria boydii EnE after near-drowning in a pond that necessitated evisceration. Endophthalmitis caused by P. boydii complex is a rare and devastating condition with poor visual outcomes. It is a condition that is difficult to treat because of its variable antifungal agent resistance. Moloney et al. [8] reported on four patients with P. boydii-associated EnE where enucleation was required in two patients and the remaining two patients had final BCVA of perception of light and no perception of light, respectively. Currently, P. boydii is recognized as the fungus most commonly implicated in invasive disease after neardrowning. These fungi are ubiquitous and are present in soil, manure, sewage, polluted water and decaying vegetation [9][10][11]. Fungal infection caused by the presence of generally harmless saprophytes in victims of near-drowning is increasingly being reported to cause serious or lethal infections, even in immunocompetent individuals [12]. Evisceration/enucleation was performed in five eyes only (7%). This is lower than what has been reported in large series. Lin et al. [6] reported that 10.42% of endophthalmitis cases underwent evisceration/ enucleation with much higher incidence of evisceration/enucleation in cases of keratitis-related endophthalmitis compared to the total endophthalmitis population [6]. All pathogens detected in those cases were either gram-negative bacilli or fungi, while Pseudomonas aeruginosa was the most commonly detected pathogen [6]. [14] that demonstrated a fivefold decrease in the endophthalmitis rate with the intracameral cefuroxime injection. Despite the marked drop in the incidence of post-cataract operation endophthalmitis, it was the leading cause of endophthalmitis in the present study, accounting for 23% of all cases of endophthalmitis and for 47% of exogenous endophthalmitis cases. This could be attributed to the fact that acute postoperative endophthalmitis in Israel is always managed as an in-patient emergency in tertiary referral centers of university-affiliated hospitals (even if the cataract surgery was performed at an ambulatory center). As the study center is the biggest tertiary referral center in Jerusalem, it possibly treated endophthalmitis cases referred from the private practice in the city and its suburbs. Endophthalmitis is a continuing concern in the era of the expanding use of intravitreal pharmacotherapy [15]. In the current study, no difference in visual outcome was observed in eyes that developed endophthalmitis after cataract operation vs eyes that developed endophthalmitis after IVI. Dar et al. [16] reported that patients with endophthalmitis following cataract surgery had better visual outcomes and were more likely to show VA improvement following treatment when compared with endophthalmitis following IVI. The authors speculated that the reason for the poorer final VA in the IVI group was not solely due to the infection-induced pathology, but rather secondary to retinal structural damage from the underlying pathology. Similarly, Simunovic et al. [17] reported that eyes with endophthalmitis following IVI had increased likelihood of a final VA of counting fingers or less and decreased likelihood of any improvement in VA following treatment in comparison to eyes with endophthalmitis following cataract operation. Also, the authors reported an increased incidence of Streptococcus spp. endophthalmitis in post-IVI cases (24.53% vs. 6.25%). McCannel [18] reported that streptococcal isolates were approximately three times more frequent after intravitreal anti-VEGF injection than after intraocular surgery. This trend, however, was not demonstrated in our study, possibly because of the use of surgical masks that help to minimize oropharyngeal droplet transmission. Coagulase-negative staphylococcus accounted for more than 40% of all positive culture results in the post-IVI group in the study by Dar et al. [16], similar to the percentage described in a meta-analysis by McCannel of 65.4% [18]. In the current study, the percentage of coagulase-negative staphylococcus was even higher, observed in 83% of the eyes following IVI. In the present series, delayed-onset endophthalmitis accounted for 20% of all cases and in 60% of the eyes with DOE, no pathogen was identified. Delayed-onset endophthalmitis represents a challenge to diagnosis since it may mimic chronic uveitis, presenting with recurrent low-grade inflammation long after intraocular surgery. It is usually secondary to Staphylococcus epidermidis and Propionibacterium acnes [19]. Revealing the offending microorganism is a difficult mission in DOE since it is generally of low-grade virulence and therefore hard to isolate [20]. Glaucoma interventional procedure was the most common intervention that preceded DOE in the index study (53% of the cases), performed at a mean of 6.8 years earlier. Similarly, Ohtomo et al. [21] reported a mean interval of 7.4 years (range 1-16 years) between trabeculectomy and the onset of bleb-related endophthalmitis. In other studies, the mean period between initial trabeculectomy and bleb-related infection was 3.1-5 years [22][23][24]. Bleb-related endophthalmitis remains a serious constant concern even after successful trabeculectomy. Anti-fibrotic agents, which are important for the maintenance of functional blebs, may lead to the formation of avascular blebs that represent a risk factor for the spread of intraocular infection through the bleb [22,23]. Sixty-three percent of patients in our cohort were treated with systemic antibiotics reflecting the debatable efficacy of intravenous antibiotics in cases of ExE. Systemic antibiotic penetration to the globe is low especially in phakic patients and is variable among the different antibiotics even in the same class (for example ceftriaxone vs ceftazidime) [25]. Arguably, it would be preferable to administer the same medication intravitreally and systemically thus reducing the elimination of the antibiotic from the vitreous [26]. But unfortunately the data of intravitreal toxicity of many antibiotics are scarce and in some medications, it originated only from animal models [27,28], which is not completely applicable to humans. Vancomycin and ceftazidime were the most commonly used intravitreal treatment in our cohort. This regimen potentially covers 100% of the bacterial infections documented but the microbiology efficacy of ceftazidime is reduced in streptococcal cases in comparison to ceftriaxone [29]. Seven isolates of different streptococci species in the present study would have been treated more efficiently had ceftriaxone been used instead of ceftazidime. As the prognosis of streptococcal endophthalmitis is poor [30], systematic evaluation of intravitreal ceftriaxone (1-2 mg/0.1 ml) [31] combined with vancomycin may be considered. The main limitation of this study is the retrospective nature of the design and the sample size not enabling further analysis on whether there was a changing trend in pathogens or their sensitivity to antibiotics over the study period. However, this study is the first panoramic report on the epidemiological, clinical and microbial profiles and antibiotic susceptibility in the Jerusalem area and the biggest study of its kind in Israel. Strategies for treatment of endophthalmitis include characterizing the epidemiologic features of patients and groups at risk, identification of the etiologic organisms and prompt initiation of broadspectrum antimicrobial therapy. Evaluation of the microbial profile and susceptibility to the frequently used antibiotics is imperative in tertiary referral settings in order to best select the empiric intravitreal antimicrobial therapy. Continuous monitoring of epidemiological data of endophthalmitis is essential for early identification of changes in the microbial ecology and susceptibility profile. The present study emphasizes the importance of adhering to the already established protocol of using vancomycin empirically as it maintained excellent effectivity and its microbial sensitivity in the present study was 100%. Authors contribution All authors contributed to collection of data, analysis, writing and reviewing the manuscript. Funding None. Data availability The corresponding author has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis as well as the decision to submit for publication. Declarations Conflict of interest The authors declare that they have no conflict of interest. Consent to publish Not relevant. Ethical approval The institutional review board of the hospital ((HMO-Hadassah Medical Organization IRB) approved the study, including waiver of informed consent for this chart review study. The study was conducted adhering to the tenets of the Declaration of Helsinki. Human or animal rights Not relevant.	2021-05-05T00:09:55.769Z	2021-03-09T00:00:00.000	{ "year": 2021, "sha1": "c4a58de56c35ffb4cfab9691b76de8080c3918cd", "oa_license": "CCBY", "oa_url": "https://www.researchsquare.com/article/rs-184174/latest.pdf", "oa_status": "GREEN", "pdf_src": "Springer", "pdf_hash": "917a9e780262e604035346d58a85434573429984", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine", "History" ] }
32773376	pes2o/s2orc	v3-fos-license	mHealth Platform and Architectures to Provide Nutritional Guidance to Children Obesity and eating disorders in children are major problems worldwide. To promote their future wellbeing, it is important to educate children and teenagers about healthy lifestyle choices. Mobile technology offers many opportunities for helping to prevent health problems; however, the sole use of stand-alone applications is insufficient to improve children’s health awareness because children require proper orientation from adults. In this paper, we present a mobile platform intended to provide nutrition guidance to children and architectures for implementing this platform in an elementary school environment. These architectures provide data capturing and user interfaces that are especially suited for children, and they permit adults to send and receive notifications and messages to improve results. KeywordsInternet of Things, mHealth, NFC, QR Codes. INTRODUCTION Obesity and excess weight have been recognized as major health problems worldwide because they increase the risk of noncommunicable diseases (NCDs). Through programs based on promoting healthy eating habits and exercise, the World Health Organization (WHO) has sought to help prevent obesity-related chronic diseases [1]- [3]. Children are also affected by excess weight and obesity. Thus, it is important that children gain increased health awareness, particularly regarding nutrition and physical activity. Today information technology offers a plethora of tools for healthcare [4], among them we consider mHealth as a very effective tool. mHealth comprises the use of mobile technology to provide health solutions. mHealth has been used to offer nutritional guidance to adults with good results. Researchers have developed a wide range of dietary and fitness applications for mobile devices, in some cases using mobile services and wireless sensor networks (WSNs). With these applications, users can improve their health when they suffer from NCDs. These applications generally require that users be aware of the nutritional facts (NFs) of food they are consuming and be able to understand the system information, which is mostly shown as graphics or raw data. In this work, we describe the development of an mHealth platform intended to be used by children aged 7 to 12 years. Working with children presents unique challenges in terms of providing adequate data-capturing methods in mobile devices and age-appropriate userfriendly interfaces. The functionality to send/receive notifications and messages from adults, including parents and/or teachers, is also important. In the platform, we used the Internet of Things (IoT) paradigm to provide the necessary features for these applications. The IoT concept [5]- [7] is based on communication among everyday objects, the ability to sense information about our environment, and the implementation of smart objects that can interact with end users and provide benefits in different application areas. The architecture of IoT has three layers [8]-[9]: 1. Perception Layer: based on technologies for device recognition, data gathering (in this case, for health-related information), and communication between sensors and mobile devices; 2. Network Layer: defines communication protocols and technologies for transmitting information captured in the perception layer; and 3. Application Layer: based on end-user requirements for application areas (in this case, focused on health prevention). We investigated how to implement IoT-based technologies for providing efficient methods to capture health-related information. We considered WSNs for capturing the physical activity information of users. This work was particularly intended to be used by children when they eat away from their homes and parents (i.e., at school). Importantly, we sought to provided architectures with an efficient communication platform between children and related adults (e.g., parents, healthcare providers, nutritionists, etc.), such that adults could remotely receive and send feedback alerts and notifications about the nutritional decisions and health status of the children. This platform would also allow adults to schedule medical appointments, if necessary, to prevent disease development. Through our investigation, we developed a prototype user interface for children to display the health information. II. RELATED WORK Emerging mobile device technologies have been used to solve health-related problems in different countries. A clear example of health prevention-related research is SapoFitness [10], which provides a mobile application for dietary evaluation in which patients can record their food PAPER MHEALTH PLATFORM AND ARCHITECTURES TO PROVIDE NUTRITIONAL GUIDANCE TO CHILDREN intake and physical activities. Medical specialists can remotely monitor the eating habits of their patients; notify patients about their progress, and offer advice about new eating habits. The main advantage of SapoFitness that can be utilized in our research is the implementation of alarm systems to notify patients about their health status. A similar approach is presented in [11]. In this work users are motivated to exercise by using a mobile game called Monster & Gold. A Bluetooth pulse oximeter is used to determine the user's heart rate. In this way the mobile game is used to train and motivate users to jog outdoors at the correct intensity. Time to Eat [12 ] is a well example on how mobile phones can be used to promote healthy habits. In this work the authors encourage children to eat well by using a mobile game. This games focus on pet care as part of a child's daily routine, which fits well with behavior change models in which repeated reinforcement is key to success. Although Time to Eat is a standalone application, it presents very good clues about developing mobile applications specially intended for children. A proposal for diabetes prevention, described in [13], implements a sensor network to obtain a patient's healthrelated information, such as blood pressure and weight. Data are captured by sensors, sent to a sensor gateway, processed, and sent to web services, where medical specialists are available to consult the patient to take preventive measures. Another related project, presented in [14], is a portable device for patients with diabetes to monitor their blood glucose levels and medications. The device also provides general knowledge about the NFs of consumed food, thus informing patients about the best food choices for their health problems. Some authors, as in [15], have investigated how to implement WSNs and mobile devices to detect suspected arrhythmias in children. Sensor networks are used to monitor ECG signals, temperature, sound, and other parameters, to verify the conditions of the children. In [16], Chapko et al. proposed a health and fitness mobile assistant to investigate the requirements and preferences of end users, presenting technologies, scenarios, and equipment to improve users' experiences of their fitness activities and, thereby, reduce the risk of chronic diseases. In our research we considered using key technologies to promote healthy nutrition and physical activity habits in children. We could observe by the studies presented above [10]- [16] that is crucial the use of wireless sensors and mobile devices to capture specific health-and physical activity-related information from users. During our preliminary research, we visited several schools in Baja California, Mexico, to obtain an overview of the activities of children and teenagers during their break times. Information regarding the children's behavior and activities was used to generate architecture proposals. We collaborated with medical specialists to identify what health information was required from users to prevent future NCDs. We determined that certain system architectures and technologies would be required for an IoT-based health-prevention platform. A. Health Information According to healthcare providers from the Instituto Mexicano del Seguro Social (IMSS) health institution in Mexico, NCDs can be prevented by promoting healthy eating habits and exercise. Accordingly, some important NFs of food need to be monitored. For our platform, we considered the following NFs: calories (ca), total fat (gt), saturated fat (gs), cholesterol (co), sodium (so), carbohydrates (ch), sugars (az), and protein (pr). We used a JSON structure for handling the NF information. JSON provides simple, rapid data handling and interoperability on mobile devices compared to XML and SGML as alternatives [17]. Figure 1 shows an example JSON structure for the NFs of a chocolate food (with fictitious data). B. IoT Architecture We used NFC tags to read food NFs through smartphones and to process and store the obtained information. After investigating NFC tag characteristics and the information that needed to be managed, we chose Mifare Classic 1K Stickers as the proper tags for this information. Users with smartphones without the NFC feature could alternatively use QR-Code tags. Users can read food NFs with their phone's camera and perform the same process for storing information. WSNs and GPS were used to capture information about the user's physical activities. However, the parameters for these sensors have yet to be defined in detail. 2. Network Layer. 3G, IPv6, and Wi-Fi protocols were implemented on this architecture to provide remote communication between mobile devices, servers, and databases containing the health information for each user. Mobile communication protocols also provided the ability to generate alerts and notifications to the users' mobile devices. Zigbee and 802.15.4 provided communication mechanisms between the mobile device and WSNs. 3. Application Layer. The main task of the application layer was to provide services to end users through the data processed from the perception layer. Software programs for mobile devices and personal computers were implemented to offer the benefits of IoT to the platform members. III. PLATFORM DESIGN Next, we defined the ideal behavior for our mobile health platform, depicted in Figure 3. 1. End User's (Child's) Module. The end user's module includes the mobile applications required by the child to track his/her food consumption by reading intelligent tags located on the food. This information is sent to the health educator or nutritionist to evaluate food consumption and provide appropriate notifications and messages. Food tags contain nutritional information and provide links to display more information about the food through augmented-reality applications. These applications are appealing to children, as they allow users to obtain information in a recreational way. Health Educator's (Nutritionist's) Module. This module consists of a Web-based system that the health educator or teacher uses to collect and analyze the user's information. Data are sent to the system with the available wireless network over the Internet. The health educator's module includes a notification system to send alarms or short motivational messages to the user when required. It also includes a messaging system that can be used to communicate with the authorized user's tutor (or parent) who needs to be informed about the user's habits. The electronic health record of each user is also included in this module. User's Tutor's (Parents') Module. This module consists of a mobile application installed on the tutor's device. Tutors can send and receive appropriate feedback and notifications when required. For instance, the health educator may inform a parent that his/her child is consuming too much sugar through a Multimedia Mobile Message (MMS) containing pictures of the items consumed. The parent then can send a warning to the child about these bad choices and take additional interventional measures at home. Physical Activity Logging Module. This module is currently being developed. It includes a set of physical devices, such as balls or other toys, equipped with a sensor that logs information about motion related to the user's activity. This information is captured as a set of points and sent to a mobile application or web-based application using WiFi. A game is being developed that will translate the amount of physical activity captured into points to keep children motivated. The ideal dynamic of the platform starts with the children continuously using the mobile apps to store NFs and physical activity (Fig. 4). Information is read with PAPER MHEALTH PLATFORM AND ARCHITECTURES TO PROVIDE NUTRITIONAL GUIDANCE TO CHILDREN automatic detection techniques, including NFC tags or QR codes for NFs and WSNs for physical activity. Data are stored locally in the mobile device, but are also sent to a web server to be analyzed by a health educator or nutritionist. Information can also be sent to parents. Nutritionists and parents can send notifications or messages to children to help them make appropriate decisions about food intake and physical activity. Figure 5 shows the platform components in our mobile health platform, which is intended to be used in elementary schools. The figure shows the interactions among all platform members. The nutritionist is able to visualize data from users and take appropriate measures to prevent health-related problems. Parents and tutors can access data and notifications to verify the health status of users. IV. PLATFORM IMPLEMENTATION AND ARCHITECTURES To provide the capability of writing intelligent tags, we developed an NFC Application Programming Interface (NFC API), using Java. The NFC API is used to store NFs with the JSON structure (Fig. 2). We used a USB NFC Card Reader/Writer (module ACR122U) and 13.56 MHz Contactless (RFID) Technology. The reader/writer supports NFC, Mifare ISO 14443 A and B cards, and FeliCa contact-less technologies. At present, we only use Mifare Classic 1K Stickers, but we provide the possibly of using QR Codes instead of NFC tags. In this case, we will use a free web service to create the QR Code and an ordinary printer to create the tags. NFs can be obtained from processed food, also a nutritionist provided us with information about food prepared in the school's cafeteria. A. Platform Architectures To develop our platform, we considered two architectures that rely on the technological characteristics of the end user's mobile device. In the first architecture (Fig. 6), the user (child) has a mobile device with the features necessary to implement IoT technologies. The user stores all of his/her health information on his/her own device. After the data are processed and servers are updated, a nutritionist can access the information for preventive health measures through notifications or medical appointments. Because most children in elementary school do not have access to mobile phones with NFC technology or equipped with a camera, we proposed a second architecture (Fig. 7), in which the person involved directly with health prevention does not have a mobile device capable of supporting IoT technologies. In this case, a "Global User" performs data entry for one or several children. The Global User may be a parent, teacher, or another student who has a mobile device capable of supporting IoT technologies. Data information can be stored in one mobile device, and the same mobile application can have several accounts to redirect data to the appropriate user. B. User Interface and Notifications To select the most suitable user interface for the mobile application to display health information to children, we performed an experiment with potential end users at two elementary schools in Mexico. There were 43 participants in the user study with ages ranging from 9 to 11 years. Of the participants, 33 own a mobile phone. Children were presented with a prototype application on a mobile device and asked to use its main features. Users were then asked to fill out a survey to verify their understanding of the screens. Note that this is a preliminary user interface; a usability study will be necessary to generate the final user interface. We used several prototypes, including showing NF with bar and pie charts and using the "Eat Well Plate" graphic. The "Eat Well Plate" ("Plato Sano" or "Plato del buen comer" in Spanish) is a graphic showing the percentage and type of food recommended to be including in a daily menu. It was defined in the Mexican official Norm NOM-043-SSA2-2005 [19], to provide unified criteria for nutritional guidelines to the Mexican population. The Eat Well Plate divides food into three food groups: fruits and vegetables, cereals and grains, and foods of animal origin (Fig. 8). In our experiments, we found that children best understood the presented information when it was displayed in the form of the Eat Well Plate, probably because schools in Mexico use the Eat Well Plate paradigm for teaching about nutrition. Figure 9 shows the main screen of the application that we created to be used by the children. This screen presents the application options. The button on the top of the screen is used to read the intelligent tag attached to a food. The button on the middle of the screen has the functionality of providing users a list of the food groups. There is also a screen describing each of the food groups included in the Eat Well Plate. The application is used to read nutritional information on a daily basis. To inform children of whether they are eating correctly, we present information using stars because this motivational technique is also used at their schools. Once a tag is read, the application uses simple images (Fig. 10) to indicate to users whether they can eat the food group that they have chosen, according to the percentage eaten at the moment. Information is also sent to a web server, where a nutritionist can analyze the data on a daily or weekly basis. Parents are informed and can send notifications to the child. We defined a set of notifications with several priorities and related communication mechanisms (Table 1). Priority refers to how quickly the notification must reach the user. The nutritionist defines the type of notification to be sent. For instance, if a child has surpassed a maximum calorie intake, he/she may decide to send a priority 0 notification (call) to inform the parents. Parents can also send a priority 1 notification (SMS) to warn the child about this. After designing the platform architectures, we generated prototypes of the capture modules on the Android O.S. and tested the prototypes with potential end users. NFC was more efficient than QR codes for capturing health information, and users preferred this technology for its simplicity of use. Concerning physical activity, we are currently working on two prototypes using a football equipped with wireless sensors. In this paper, we have presented the necessary architectures to develop an IoT-based platform for the prevention of obesity-and poor nutrition-related NCDs. Comparing our results with current health-prevention systems; we conclude that our technologies provide more efficient capture mechanisms to end users. As a key to the prevention of health-related problems, our platform also uses an alarm system. Because it is specially designed for children and teenagers, the user interface will improve the end user's experience with the mobile health system. Future improvements will include the development of prototypes based on the proposed architectures, the implementation of different WSNs, and the verification and documentation of the platform's behavior with potential users.	2018-01-23T22:38:23.035Z	2013-10-12T00:00:00.000	{ "year": 2013, "sha1": "d3355cdae8650a2c46694468f5f3b9da186d9867", "oa_license": "CCBY", "oa_url": "https://doi.org/10.3991/ijim.v7i4.3083", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "2ee622897aca60cd5eaadadba9c3e8fefaf7d6c0", "s2fieldsofstudy": [ "Education", "Medicine" ], "extfieldsofstudy": [ "Computer Science" ] }
248813513	pes2o/s2orc	v3-fos-license	Prevalence of Physical Activity Among Healthcare Students in King Saud University, Riyadh, Saudi Arabia. An Observational Study Background and Objective Physical inactivity is 1 of the most important cardiovascular risk factors causing higher morbidity and mortality rate worldwide. Thus, the promotion of physical activity (PA) is 1 of the fundamental pillars in the primary prevention of heart disease. Therefore, this study aimed to assess the prevalence of Physical Activity among healthcare Students in the Riyadh region, Saudi Arabia. Methods An observational, descriptive, cross-sectional study using paper-based printed questionnaires was conducted among Emergency medical students (EMS) from the Prince Sultan College for Emergency Medical Services King Saud University. To evaluate the physical activity, Global Physical Activity Questionnaire (16-items) developed by the World health organization was used. The questionnaires talk about activity at work (6 items), travel to and from places (3-items), recreational activities (6- items), and sedentary behavior (1-item). Both binary answers (Yes\No) and continuous scales were used to collect the data. Data were descriptively analyzed using a statistical package for social science version 26 (SPSS). Results The prevalence of PA among students was 69.9% (n = 123). The most common type of physical activity was walking, accounted for 51.7% (n = 91), followed by Gym/bodybuilding 25% (n = 44), running 24.4% (n = 43). A total of 7.16 mean hours (SD = 11.23) were spent by students doing vigorous-intensity activities on a typical day, while 4.93(SD = 8.84) of mean hours were spent on moderate-intensity activities. Additionally, more than half of the students 55.7% (n = 98) performed vigorous-intensity sports, fitness, or recreational activities that cause large increases in breathing or heart rate (for at least 10 minutes continuously). The mean time spent by the students in this study performing vigorous-intensity sports, and fitness activity were 4.60(SD = 10.22) mean hours per day. The sedentary behavior, like constant sitting, was reported to be 7.33 (SD = 4.24) mean hours per day. Conclusion In conclusion, about 3 in ten EMS students are physically inactive, which could be due to study-related actions. Future research among undergraduates should focus on approaches to increase PA is needed. Introduction Physical inactivity is a raising concept health issue linked to various acute and chronic diseases and lifestyle conditions which include obesity, diabetes, cardiovascular disease, and gastrointestinal problems which include colon cancers. 1,2 It is assumed that relatively a high prevalence of disease was mainly due to a lack of physical activity (PA) among Individuals. 2,3 According to recent findings from the World Health Organization, worldwide 1 in 4 adults does not fulfill PA requirements. 4,5 Furthermore WHO recommended at least 150-300 minutes of moderate-intensity aerobic PA for adults or 75-150 minutes of vigorous-intensity PA or a combination of it is needed throughout the week. 5 Although the previous report suggested that lack of PA is directly responsible for 5.3 million deaths annually. 6 In addition, WHO, reported that Physical inactivity is 1 of the ten major causes of mortality in the world and imposes a significant financial burden, estimated to be 11% of total healthcare spending in the United States, per year. 7 In Saudi Arabia, a recent report estimated the health care costs associated with lack of physical activity were found to be $869,019, representing 1.71% of the total healthcare costs. 8 Physical activity is defined as any bodily movement that requires energy expenditure and is performed by skeletal muscles. Furthermore, energy expenditure associated with PA is measured in Kilocalories. 5 Physical activity promotes increased oxygen consumption in comparison to the relaxing stage of the human body, to meet the required energy during this procedure, the skeletal system of the body uses its storage of free fatty acids and triglycerides, thereby the supply of blood to the whole body are remarkably well maintained, which in turn leads to the healthy body after the Physical activity. 9 WHO reported multiple benefits of regular physical activity, which includes improvement in muscle movements of the body and cardiorespiratory fitness, helps in maintaining healthy bones and skeleton system. 5 Furthermore, PA activity also helps in reducing the risk of heart attaches and cardiovascular diseases. 5,[10][11][12][13] Other studies reported improvement in the mental ability of the individuals and overall quality of life. 10 The individuals with low PA were more prone to suffer health problems than those who are active physically. Physical inactivity is significantly associated with the high risk of cardiovascular disease, endocrine disease, elevated levels of cholesterols and triglycerides, and risk of development of various cancers and, increased risk of weight gain. 5,[9][10][11][12][13] Despite the high prevalence of smoking and unhealthy eating habits, and an increasingly sedentary lifestyle among adolescents and university students, activities involving sitting, reclining, or lying down is more prevalent. 14 Although studies also highlighted that university students were present with a relatively higher prevalence of overweight and obesity compared to the general population. 15 Additionally, there was evidence that undergraduates were reported to have a higher consumption of unhealthy drinks and fast food, and dietary supplements to cope with the stress associated with their lifestyle and academic routine. [16][17][18][19][20] Multiple previous studies demonstrated level of PA was not sufficient among students throughout the world. 21,22 In Saudi Arabia, recent literature found that majority of the adolescents including college students were not active to meet the required guidelines for moderate PA. 23 The prevalence of physical inactivity was reported differently in various studies, for instance, it was 34.8% among health care professionals in Saudi Arabia, 24 the highest prevalence of 82.6% was reported among the public in Saudi Arabia 25 while it was 49% among Malaysian medical students. 26 The prevalence of physical activity was measured using the WHO stepwise questionnaire. 24,25 Overall, PA among individuals is crucial for maintaining a healthy life. Despite the wealth of literature about PA behaviors in the general public in Saudi Arabia and other countries. [23][24][25] There is a scarcity of research on the health care' students toward physical activity. The present study aims to evaluate the prevalence of physical activity, among pharmacy students in Riyadh, Saudi Arabia. Study Design, Participants, and Settings A cross-sectional questionnaire-based study was conducted between December 2021 to February 2022, among emergency medical service (EMS) students from the Prince Sultan College for Emergency Medical Services in King Saud University, Riyadh, Saudi Arabia. The targeted population included EMS students aged ≥18 years, speaking Arabic or English, and willing to participate in the study by signing the consent. The participants were assured that the data would be used only for research and would be maintained confidential throughout the study. Others who do not match the inclusion criteria were excluded. Questionnaire Design The questionnaires used for this study were prepared after an extensive review of similar studies published elsewhere. 25 (Appendix 1) Furthermore, the WHO stepwise questionnaire was also used to assess the PA. 26 The questionnaire was divided into 2 sections. The first section included demographic items that assessed participants' characteristics like age, gender, and information about their PA status. The second section contained 16 items divided into 4 sub-sections. Subsection 1 consisted of 6-items about activity at work., The second subsection talks about travel to and from places (3-items), recreational activities 6-items, and sedentary behavior 1-item. Both binary answers (Yes\No) and continuous scales were used to collect the data. In this study, the prevalence of physical activity (walking, cycling, bodybuilding, running, or any sports-related activity) was determined, by individuals or students who practice any kind of physical activity on weekly basis. The questionnaire was translated into Arabic by a certified Arabic speaker. The questionnaire was evaluated by a research team, to ensure the readability and accuracy of the content. Furthermore, a pilot study was conducted among a randomly selected sample of EMS students (n = 20). The results of the pilot study reveal the length of the time of the survey, which was estimated to be 7 minutes. Additionally, the result of the pilot study was not included in the final analysis. The reliability of the questionnaire was determined using Cronbach's Alpha value of (.85), indicating that questionnaire was reliable to carry out the study. Data Collection The data were collected using the convenience sampling method. The final study questionnaire was distributed by visiting personally each year the students, who were currently enrolled and present at the time of data collection. Paperbased printed questionnaires were used to collect the data. To collect the maximum number of responses students were followed by a team of researchers (1 researcher and a chief investigator). To avoid response duplications, participants were instructed to fill out the questionnaire once despite the method through which they received it. Data Analysis The data were analyzed using the Statistical Package for the Social Sciences (SPSS) version 26 for Windows (SPSS Inc, Chicago, Illinois). Descriptive statistic was used to summarize the demographic characteristics. For the continuous data values were presented in the form of mean and standard deviation. Results A total of 210 students completed the questionnaire, of which 176 completely answered the questionnaire giving a response rate of 83.8%. Of which 98.3% (n=173) were aged between 18 to <25 as shown in Table 1. The Second Year Students group was predominant with 38.1%, followed by the third year's = 34.7% (n=61). About slightly more than one-third of the students (36.4%) reported a family income of >5000 Saudi Arabian riyals (SAR), while 25.0% of them reported 11 000-20 000. The prevalence of PA among all students was 69.9% (n=123). Detailed information on the demographic characteristics and smoking status of the respondents were shown in Table 1. When we asked about the Reasons for physical inactivity among them, the majority of the students reported a lack of time 47.2% (n=83), followed by an unwillingness to practice physical activity 21.6% (n=38) and injury10.2% (n=18). Furthermore, detailed information on the Reasons for physical activity of the respondents was shown in Figure 1. In this study 7.16hours (SD=11.23) of the meantime were spent by the health care students doing vigorous-intensity activities at work on a typical day (Median (2)), while 4.93(SD=8.84) of mean hours were spent on moderateintensity activities at work ( Figure 2). Slightly less than half of the students 40.6% (n=71) used bicycles for at least 10minutes continuously to get to and from places. Additionally, more than half of the students 55.7% (n=98) performed vigorous-intensity sports, fitness, or recreational activities that cause large increases in breathing or heart rate like [for at least 10 minutes continuously. The mean time spent by the students in this study performing vigorousintensity sports, and fitness activity were 4.60(SD=10.22) mean hours per day. Sedentary behavior like constant sitting was reported to be 7.33 (SD=4.24) mean hours per day ( Figure 2) (Table 3). In this study, one-third of the students were inactive physically. One-third of the students 30.3% (n=53) did vigorous-intensity activity only 1 day a week, while 24% (n=42) of them followed it for 2 days a week. The moderateintensity activity was followed by 21.8% of the students for 2 days in a week, while vigorous-intensity sports, fitness, or recreational activity was followed by 3 days in a week 15.4% of the students. Detailed information on the days spent doing different types of PA in a typical week was shown in Table 4. Discussions The physical activity of the body has been associated with significant benefits to the individuals and was widely established in the previous literature, [25][26][27][28][29] and it was encouraging to discover in our study that the majority (69.9%) of health care students are regularly active physically. This study results were higher than the previous study by Naim et al among medical and non-medical students in Malaysia (51%), 30 29 The prevalence of PA may differ from 1 study to another and may be influenced by several factors including the study method, types of respondents, and demographics of the subjects. However, in many developed and developing countries. Physical inactivity is increasing and is reported to be a common practice among students, due to the increased sedentary lifestyle. Although in the current study the work-related vigorousintensity activity was found to be 54%. Similar results were reported by Marques-Sule et al among health care professionals who reported most of the participants were physically active at their workplace, during their spare time. 29 Similarly, a previous Study by Al-Zalabani et al, in Saudi Arabia reported 18.3% of work-related PA. 33 However, according to a recent systematic review, the prevalence of physical inactivity among young Saudis was 71%. 23 Lack of PA is significantly associated with obesity and hypertension, hyperlipidemia, and glucose intolerance are the first harmful effects of obesity to appear in a population in a socioeconomic transition state, while coronary heart disease and long-term diabetes complications, such as kidney failure, appear later. Although previous studies concluded that health care professionals were found to be physically inactive in comparison to other populations. 32 In this study walking, running, bodybuilding and playing games were the types of PA performed by Saudi health care students. However, the most common justification to avoid physical activities was lack of time. Similar results were reported by Awadalla et al among health college students, by Al Reshidi in 2016 among residents. 34,35 Although Wattanapisit et al reported lack of social support was another potential barrier that influence the PA behavior of the students. 29 Additionally, studies found that the type of PA activity among students was mainly supported by environmental, social, psychosocial, and cultural characteristics. 29,36 Cumulatively and not surprisingly in this study, the average amount of time spent constantly sitting was 7.33 hours. Although the World Health Organization (WHO) recommends that individuals engage in at least 75-150 minutes of vigorous-intensity aerobic PA per week or an equivalent combination of moderate-and vigorous-intensity activity throughout the week. 5 In the current findings, 7.16 mean hours were spent by health care students doing vigorous-intensity activities at work on a typical day. While moderate-intensity sports, fitness, or activity was reported to be 3.03 hours. Unsurprisingly, undergraduates were less engaged than other people. This could be because that is the current study's healthcare students spent a lot of time in hospitals conducting clinical rotations or practical cessations, and their free time was dominated by study-related activities. Physical exercise among future healthcare practitioners would assist them in role modeling and promoting PA among their patients. Knowing the scope of physical activity would aid in the development of policies to promote it regularly among individuals living in Saudi Arabia and other countries, and keeping sufficient knowledge and awareness of PA in current study participants who are regarded as future professionals would aid in providing counseling services to their patients. In terms of the importance and usefulness of PA methods among participants, despite being physically active, the majority of students did not fully understand the importance and effectiveness of PA methods according to WHO recommendations. Additionally, it is important to incorporate education about PA programs or raise awareness of PA into their academic lives to improve their knowledge of PA methods. According to a previous study in Saudi Arabia, increased cardiovascular diseases and obesity, are other interventions to reduce physical inactivity prevalence in Saudi Arabia and other countries. Our research has a few limitations. First, the design of the research was a based-cross sectional self-reported survey, potentially rendering our results less reliable. However, because the survey was anonymous and completely voluntary, 1 can assume that PA status was reliably captured. Secondly, the study was conducted on a single university's students limited to EMS, hence results cannot be generalized to all health care students in Saudi Arabia. Lastly, this study was conducted without determining the power analysis for the sample size estimation. Conclusion Our study highlights an increasing trend of Physical inactivity with low levels of minimum PA among health college students living in the capital of Saudi Arabia, Riyadh. More importantly, the prevalence of physical inactivity is on a rise compared with both national and international studies, which may potentially associate with negative consequences on public health. Therefore, we advocate the implementation of programs or workshops that educate students about the harmful outcomes surrounding physical inactivity and their consequences on overall health. An Inter-professional approach to education that addresses the importance of using PA could be the most practical way to address such a complex issue. Table 5.	2022-05-17T06:23:52.380Z	2022-01-01T00:00:00.000	{ "year": 2022, "sha1": "d26543538e13d963b121393a19bd1fd454a7f75a", "oa_license": "CCBYNC", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "590194c513e9f62a2948d7302f3eab20dd6bbf24", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
252738353	pes2o/s2orc	v3-fos-license	Chromatin regulators in the TBX1 network confer risk for conotruncal heart defects in 22q11.2DS Congenital heart disease (CHD) affecting the conotruncal region of the heart, occurs in 40–50% of patients with 22q11.2 deletion syndrome (22q11.2DS). This syndrome is a rare disorder with relative genetic homogeneity that can facilitate identification of genetic modifiers. Haploinsufficiency of TBX1, encoding a T-box transcription factor, is one of the main genes responsible for the etiology of the syndrome. We suggest that genetic modifiers of conotruncal defects in patients with 22q11.2DS may be in the TBX1 gene network. To identify genetic modifiers, we analyzed rare, predicted damaging variants in whole genome sequence of 456 cases with conotruncal defects and 537 controls, with 22q11.2DS. We then performed gene set approaches and identified chromatin regulatory genes as modifiers. Chromatin genes with recurrent damaging variants include EP400, KAT6A, KMT2C, KMT2D, NSD1, CHD7 and PHF21A. In total, we identified 37 chromatin regulatory genes, that may increase risk for conotruncal heart defects in 8.5% of 22q11.2DS cases. Many of these genes were identified as risk factors for sporadic CHD in the general population. These genes are co-expressed in cardiac progenitor cells with TBX1, suggesting that they may be in the same genetic network. The genes KAT6A, KMT2C, CHD7 and EZH2, have been previously shown to genetically interact with TBX1 in mouse models. Our findings indicate that disturbance of chromatin regulatory genes impact the TBX1 gene network serving as genetic modifiers of 22q11.2DS and sporadic CHD, suggesting that there are some shared mechanisms involving the TBX1 gene network in the etiology of CHD. INTRODUCTION Congenital heart disease (CHD) occurs sporadically in approximately 1% of the general population resulting in significant morbidity and mortality. A subset has syndromic causes with known genetic etiologies. The 22q11.2 deletion syndrome (22q11.2DS; also named DiGeorge syndrome or velo-cardio-facial syndrome) is an example of a genetic syndrome in which the majority have CHD. It is estimated that 40-50% of individuals with 22q11.2DS have conotruncal heart defects (CTDs) affecting the formation of the cardiac outflow tract (OFT) and/or aortic arch 1 . TBX1 encodes a T-box transcription factor mapping to the hemizygously deleted region on chromosome 22q11. 2 and it has been shown to be largely responsible for the syndrome's cardiac phenotype [2][3][4] . Further, mutation of one allele of TBX1, without a 22q11.2 deletion, is responsible for CTDs, confirming its importance as a disease gene 5 . Inactivation of one allele of Tbx1 in mice results in mild aortic arch anomalies while inactivation of both alleles results in neonatal lethality with a persistent truncus arteriosus, at complete penetrance [2][3][4]6 . Although haploinsufficiency of TBX1 has a major impact in the etiology of disease, it cannot fully explain variation in the frequency of occurrence of CTDs and therefore it is likely that some of this is due to the existence of genetic modifiers. Our goal is to identify genes that modify the effects of the 22q11.2 deletion as related to the haploinsufficiency of TBX1. We focused on rare coding and splicing variants that might affect protein function, using similar approaches as genetic studies of sporadic CHD 7,8 . One possibility is that there are damaging coding/splicing variants on the remaining allele of TBX1 or other genes in the hemizygous 22q11.2 region that can alter risk of CTDs, however mutations were not detected 9,10 . Therefore, we decided to investigate potentially damaging rare coding/splicing variants in the rest of the genome that may serve as modifiers of 22q11.2DS in the TBX1 genetic network. In this report, we analyzed whole genome sequence (WGS) of 1182 subjects with 22q11.2DS, where we focused upon 456 cases with CTDs versus 537 with a normal heart serving as controls to identify variants that affect protein function. Rare, predicted most damaging variants were identified and are termed, MDRV. We used an unbiased method to identify recurrently affected genes, focusing on gene functional categories. From this analysis, chromatin regulatory genes were identified in cases but not controls. This was followed by a gene set approach to investigate 19 gene sets chosen from genes that are constrained, essential to cell function, genes that are likely to have features of haploinsufficient genes as well as gene sets chosen from the Pediatric Cardiac Genomics Consortium (PCGC) for studies of sporadic CHD in the general population. We then used two different statistical approaches for analysis. We used an over-representation method and a weighted gene set method based upon gene expression, on genes affected in one or more subject. We then compared our findings to that of sporadic CHD in the general population as a validation step and examined expression of the genes in mouse embryonic transcriptomic data relevant to heart development. We identified specific chromatin regulatory genes including histone lysine acetyltransferases and lysine methyltransferases, of which some are in the TBX1 molecular network, and secondly, that they strongly overlap with genes found as risk factors for sporadic CHD in the general population. WGS quality control and principal component analysis We obtained WGS from 1182 de-identified subjects with 22q11.2DS from the International 22q11.2 Brain and Behavior Consortium (IBBC) [10][11][12][13] . We then performed quality control measures and after this, there were a total of over 21 million variants that remained ( Supplementary Fig. 1). The number of single nucleotide variants and indels with one or more alternative alleles were determined ( Supplementary Fig. 2a). In this cohort, indels comprised 8.62% of the total number of variants ( Supplementary Fig. 2a). Most of the indels were within 10 bp in size ( Supplementary Fig. 2b). The alternate allele frequency distribution for the cohort versus the number of variants were identified and most of the variants occurred in only one subject ( Supplementary Fig. 2c). Principal component analysis (PCA) was then performed to identify the ancestry groups among the 1182 subjects with 22q11.2DS (Supplementary Fig. 3; Supplementary Table 1). The subjects were distributed in Caucasian, African Descent and Hispanic ancestry groups, using the HapMap population ( Supplementary Fig. 3a, b). Study design To identify genetic modifiers, we compared 456 cases with CTDs (22q11.2DS-CTDs) and 537 controls without CHD, all with 22q11.2DS ( Fig. 1a; Supplementary Table 1). We excluded those with isolated ventricular septal or atrial septal defects because their developmental origins are more complex than for CTDs. We focused on rare variants that might affect the function of proteins along with haploinsufficiency of genes in the 22q11.2 region, during cardiac development. Our study design is illustrated in the flow diagram in Fig. 1b. We applied PEMapper/PECaller 14 to identify variants and we then validated with GATK 15 ( Fig. 1b; Supplementary Fig. 4). The purpose of using the two different variant calling algorithms was to rule out false positive variants. Rare variants had an alternative allele frequency (AAF) in genomAD of less than 1% (Fig. 1b). Comprehensive annotation software and algorithms were used (see Methods) to identify 55,379 variants including LoF, splicing and putative damaging missense variants (Supplementary Table 2; Supplementary Fig. 4). Additional software and algorithms were used for further filtering (phastCons > = 0.5, for evolutionarily conserved variants; CADD > = 10 for deleteriousness of variants, CCRS > = 80 for constrained coding regions; Supplementary Fig. 4) to identify 1861 highconfidence, predicted most damaging rare coding/splicing variants (MDRV, Fig. 1b Supplementary Fig. 6b). Tbx1 is expressed in the pharyngeal apparatus (PA; arches 2-6) and in cells that migrate to the cardiac OFT and aortic arch arteries between mouse embryonic day (E)8-10.5 16 . To filter for genes most likely to be relevant to progenitor populations of the heart and TBX1, we generated RNA-seq data to determine the expression level of genes in the PA, OFT with right ventricle (OFT + RV), and/or left ventricle with atria (LV+atria) of mouse embryos at E10.5 (Fig. 1c). Of the protein coding genes in the mouse genome in each tissue, we considered the top 35% as expressed (with > = 25 reads per million; Supplementary Table 5) and referred to them as cardiac progenitor cells. Chromatin regulatory genes identified with recurrent MDRVs in 22q11.2DS-CTD cases We selected recurrently affected genes with MDRVs in at least two cases and no controls to help avoid possible random sequencing artifacts. A similar criterion was applied to the control group (two or more controls, no cases; Methods). In the 456, 22q11.2DS-CTD cases, there were 413 genes with MDRVs, 31 of which were recurrently affected. In 537 controls, there were 466 genes with MDRVs, and 38 were recurrently affected. We found 12 of 31 recurrent genes in cases (Table 1; 5.7% of cases), and 13 of 38 recurrent genes in controls, were expressed in mouse embryonic cardiac progenitors (Supplementary Table 5). In total, we did not find more genes affected in cases versus controls with 22q11.2DS. We next wanted to focus on the functional categories of genes identified in cases versus controls that were expressed in cardiac progenitor cells. Strikingly, half of the 12 recurrent genes with expression in cardiac progenitor cells in 22q11.2DS-CTD cases were chromatin regulatory genes: NSD1, KAT6A, KMT2D, PHF21A, EP400 and KMT2C, of which some are associated with known syndromes that include CHD, supporting their candidacy (Table 1). Table 2 lists the Fig. 1 The 22q11.2 deletion syndrome cohort and study design. a Pie chart of intracardiac and aortic arch phenotypes. Control (gray, no significant heart defect); CTD (conotruncal heart defect, blue); ASD alone (isolated atrial septal defect but no other heart or aortic arch defects, light blue); VSD alone (isolated ventricular septal defect but no other heart or aortic arch defects, light blue). Pie chart includes controls (gray) versus CTD cases with phenotypes including: TOF (tetralogy of Fallot, light blue), RAA (right sided aortic arch, orange), IAAB (interrupted aortic arch type B, green), PTA (persistent truncus arteriosus, yellow), PS/PA (pulmonary stenosis and/or pulmonic atresia, blue) and other aortic arch defects such as abnormal origin of the right or left subclavian artery, alone (light green). b Schematic representation of the case-control study design using WGS. Variants were identified using PEMapper/Caller and validated by GATK. Only shared variants between both pipelines were used. Following quality control measures of the raw WGS data, variant annotation was performed to identify rare (< 1%) predicted LoF (loss of function), damaging splicing and damaging missense variants followed by filtering-based annotation on phastCons (conservation), CADD and CCRS (constrained coding regions) scores to identify MDRVs. Then gene set analyses were performed including over-representation (ORA) and weighted gene set based tests. STRING analysis was performed to identify potential biological network interactions. c Lateral side of mouse embryo at E10.5 with outline of tissues used for bulk RNA-sequencing (pharyngeal arches 2-6, PA2-6, blue; outflow tract and right ventricle, OFT + RV, green; left ventricle and atria, LV+atria, yellow). detailed annotation of the MDRVs identified in the six recurrently affected chromatin regulatory genes, as well as the associated cardiac phenotypes. The most frequent cardiac phenotype was tetralogy of Fallot followed in frequency by right sided aortic arch (Table 2). Moreover, there were MDRVs in KMT2C in two additional 22q11.2DS participants (not included here as CTD cases or as controls), one with a ventricular septal defect (VSD; with no other heart or aortic arch anomalies) and one with an atrial septal defect (ASD; with no other heart or aortic arch anomalies; Tables 1 and 2). Although there was no increase in the presence of MDRVs, thus, no increase in burden, we found a specific enrichment of a particular functional class of genes. We found that chromatin regulatory genes were greatly enriched among all possible functional categories of genes in the genome, whereas no single functional category of genes was enriched in controls. Over-representation analysis of recurrent genes identifies chromatin regulatory genes in 22q11.2-CTD cases but not controls Considering both the rarity of high-confidence MDRVs and relatively limited number of samples included, we next used a gene set based approach to test the aggregated effect of MDRVs in multiple functionally or conceptually connected genes in 19 gene sets, in cases and controls (Fig. 1b). The 19 gene sets were derived from three different sources of which had between 100 and 1000 genes per set (Supplementary Table 6). There was a total of 5403 genes in the 19 gene sets, of which 347 had one or more MDRVs. This number of genes per set ensured sufficient power for our statistical tests. One source included constrained, haploinsufficiency, and essential gene sets that are under strong purifying selection 17 , tend to play important roles in protein interaction networks 18,19 , and are highly enriched for disease genes 20 . The second source was chromatin regulatory gene sets because some of this functional category was identified in this study (Table 1). Additionally, de novo mutations in chromatin regulatory genes were found in individuals with sporadic CHD by the Pediatric Cardiac Genomics Consortium (PCGC) 21 . One gene set consisted of 866 genes (All Chromatin Regulatory Genes, combined from several sources; Supplementary Table 7) and the other consisted of 274 genes and was a smaller subset (REACTOME term Chromatin Modifying Enzymes, Supplementary Tables 6 and 7). The third source comprised 14 gene sets (116 to 791 genes per set), derived from genetic studies of sporadic CHD by the PCGC, including genes found with mutations in one or more subjects 7 . These 19 gene sets, though curated using differing concepts, are not mutually exclusive, as shown in the Venn plots in Supplementary Fig. 7, indicating various degrees of overlap among different gene sets. We then performed an over-representation analysis (ORA) of the recurrently affected genes on 19 gene sets (Fig. 2). We found that five of the six recurrent chromatin regulatory genes described above (Table 1; not EP400) contributed to findings for the Constrained Genes set with borderline significance (n = 968; Fig. 2, panel 1, P = 5.28 × 10 −3 ). The chromatin gene sets showed significant over-representation, with all six genes shown in Table 2, either when filtered by expression in cardiac progenitor cells or not filtered (Fig. 2, panels 1 and 2; Supplementary Table 8). Further, four of these six genes contributed to findings for the Candidate CHD Genes set created by the PCGC (not EP400 or PHF21A; n = 402; Fig. 2, panel 2). None of the other gene sets were overrepresented nor were the three non-cardiac gene sets used as controls ( Fig. 2; Supplementary Table 8). In the control 22q11.2DS samples, there was no enrichment of any gene set by ORA for the recurrently affected genes, regardless of developmental cardiac expression gene filtering (all P > 0.05 before correction; Fig. 2, panels 3 and 4), suggesting stochastic effects of MDRVs in other genes in controls. nonduplicated 24 26 Genes are listed alphabetically among the six chromatin genes in bold font. Of note, KMT2C is also affected in one ASD patient and one VSD patient, with no other heart or aortic arch defects; both carrying same variant rs145848316 as described in Table 2. MDRV most damaging rare variants. AD Autosomal dominant. Y. Zhao et al. AWS,Domain, Associated With SET; 1 denotes male and 2 female, in sex column; All MDRVs (most damaging rare variants) reside in Consensus coding sequence regions (CCDS). GnomAD v3.1.2 (CEU+Latino/ Admixed +African/African American). Rs145848316 in KMT2C was also identified in one isolated ASD patient and one isolated VSD patient, see Table 1. None of the other genes were identified for isolated ASD and isolated VSD. ACMG classification is derived from Franklin Genoox (https://franklin.genoox.com/clinical-db/home), VUS Variant of unknown significance. Y. Zhao et al. Weighted gene set test expands chromatin regulatory genes identified in 22q11.2-CTD cases We next employed a weighted gene set based test to evaluate whether high-confidence MDRV burden was significantly concentrated within any of the 19 gene sets. For this analysis, the mean expression level in cardiac progenitor cells from RNA-seq analysis of E10.5 mouse embryos was used. Again, we found chromatin regulatory genes contributing to 22q11.2DS-CTDs but identified a broader set of genes using this approach. The weighted gene set approach can reduce noise/signal ratio by prioritizing genes by their functional importance, and more specifically, by expression Fig. 2 Over-representation analysis (ORA) of recurrently affected genes identifies chromatin regulatory genes contributing risk to CTDs in 22q11.2DS. Three different sources of gene sets totaling 19 are indicated by color below the bar graph (gene sets used by the PCGC to investigate sporadic CHD 14 are indicated by black box, as well as in lilac and in gray). The first bar on the left in each panel shows the total number of recurrently affected genes (n) among all affected genes (N). The rest of the bars indicate the number of recurrently affected genes within each gene set (k) versus the final number of affected genes with MDRVs (most damaging rare variants) for each gene set (M) as indicated (see Methods for more details). The top two bar graphs show ORA results without filtering by gene expression levels in CTD cases (red) and with filtering (dark red) followed by the same for controls (green and dark green, respectively). The numbers in some of the bars denote the number of recurrently affected genes contained within the specific gene set / the total number of affected genes in this gene set. The gene set analyses were corrected for multiple testing by false discovery rare (FDR). Red asterisks denote significance after FDR correction; blue asterisk denotes borderline significance (P = 0.057). level during cardiac development, since genes in a gene set are not expected to contribute to disease equally. Genes affected with one or more MDRVs in one or more subjects, were included, thereby making it possible to expand the number of potential modifier genes. One gene set, Chromatin Regulatory Enzymes (274 genes), had significantly excess burden of MDRVs in 22q11.2DS-CTDs versus controls after false discovery rate (FDR) correction, with 24 identified MDRVs in cases and 14 MDRVs in controls (P = 5.00 × 10 −4 , Fig. 3, Supplementary Table 9); there was no significant enrichment of MDRVs in any of the three noncardiac tissue gene sets (P > 0.05 before FDR correction). The MDRVs that each affect one or more chromatin gene is either ultra-rare (n = 12, AAF < 1.76 × 10 −4 ) or novel (n = 30) in gnomAD (Supplementary Table 10). When taken together with ORA and weighted gene set approaches, we identified 42 MDRVs in 37 chromatin genes that occurred in 39, 22q11.2DS-CTD cases, accounting for 8.5% of 22q11.2DS-CTD cases in total. We found that 18 of the 37 chromatin genes are associated with known genetic syndromes that include CHD, supporting their candidacy (Supplementary Table 10). Overlap of chromatin regulatory genes between 22q11.2DS-CTDs with sporadic CHD, but differences in pathogenicity of variants We then compared the genes identified in 22q11.2DS-CTDs with chromatin regulatory genes found in cases of sporadic CHD from the PCGC. Sporadic CHD is defined by the PCGC as isolated cases in which neither parent is affected. Individuals with known syndromes, such as 22q11.2DS, were excluded, however subjects with isolated CHD or CHD that co-occurred with neurodevelopmental deficits or additional congenital anomalies were included 21 . Most of the chromatin regulatory genes found in studies by the PCGC had de novo mutations and most occurred in sporadic CHD with neurodevelopmental disorders or extracardiac features 21 . We cataloged genes that overlapped and found that 13 of the chromatin regulatory genes with MDRVs in the 22q11.2DS-CTD cases were identified to have de novo mutations (all categories: protein truncating, splicing, missense) in one or more subjects with sporadic CHD from the PCGC (total of 90 chromatin regulatory genes with MDRVs among 2871 CHD probands; Fig. 4a, b; Supplementary Table 11) 7,21,22 . In an independent WES study of sporadic CHD to identify genetic risk factors, Sifrim and colleagues examined sequence variants in similarly categorized individuals with sporadic CHD 23 . Nine genes in 22q11.2DS-CTD cases were found among 65 genes identified with de novo mutations in CHD with neurodevelopmental disorders or extracardiac features as shown in a Venn diagram (Fig. 4a) and UpSet plot (Fig. 4b) 23,24 (610 subjects, syndromic-CHD; S-CHD cases; Fig. 4a, b). In an integrated analysis 23 , four chromatin genes in 22q11.2DS-CTD cases were identified among 16 found (16 de novo and inherited variants were found at the highest tier of significance among 1891 probands with/out neurodevelopmental disorders or extracardiac features in S-CHD vs nonsyndromic, NS-CHD; Fig. 4a, b). When taken together, 14 genes were shared among the 22q11.2DS-CTDs, PCGC and Sifrim et al. studies (CHD7, KAT6A, KMT2C, KMT2D, NSD1, SMAD4, TRRAP, EP400, IPO9, BRPF3, DNMT3A, HLTF, KDM4B, KMT2E, Supplementary Table 11), thereby providing further support of their role in cardiac development and disease. This finding is particularly compelling when considering that the identification of rare variants was performed using different patient cohorts (22q11.2DS-CTDs, sporadic CHD in general population by the PCGC 7,21,22 and by Sifrim et al. 23 , study design (case-control, trios, and both trios and singletons, respectively), variant types (unknown inheritance for 22q11.2DS-CTDs, de novo and inherited recessive variants for sporadic CHD studies) and partially overlapping variant annotation pipelines/statistical methods. Further, of the 14 genes among the three studies, eight of ten that are causative of genetic syndromes have CHD as a notable feature [25][26][27][28][29][30] . Despite differences outlined above, overall, the focus of these studies was on rare variants associated with CHD that were deemed to be pathogenic, providing some similarities as well with the 22q11.2DS study (Supplementary Table 12). We next wanted to compare variants with pathogenicity identified in the clinical literature. Most of the MDRVs in 22q11.2DS-CTD cases were missense changes predicted to be damaging (Fig. 4c). From examination of clinical databases, 39 of 42 were considered variants of unknown significance (VUS; Table 2 and Supplementary Table 10). We then wanted to ask whether the types of variants in chromatin regulatory genes found in 22q11.2-CTDs might be similar to those found in studies of sporadic CHD. In contrast to 22q11.2-CTDs, protein truncating variants (PTVs; loss of function, LoF) were found more frequently in sporadic CHD by the PCGC (Fig. 4c) 21,22,31 . This was similar for studies by Sifrim et al. 23 (Fig. 4c). Further, many of the variants identified in studies of sporadic CHD, in particular for chromatin regulatory genes, were de novo heterozygous mutations, but the variants we identified were of unknown inheritance. This suggests that variants in chromatin regulatory genes in 22q11.2DS-CTDs may affect gene function but are not disease causing on their own, thereby serving as modifiers. Biological network connections between chromatin regulatory genes and TBX1 To understand the biology of the genes, and with respect to TBX1, we generated a functional protein network using STRING software (https://string-db.org: Fig. 5a). The network consists of the 37 chromatin regulatory genes we identified plus two additional genes, CHD7 and ATAD2B, where we found MDRVs affecting more cases than controls in this study (two versus one, three versus two). As expected, there is an appreciable amount of functional coherence of these genes (P = 5.03 × 10 −14 ). These chromatin regulatory genes have functions as histone lysine acetyltransferases and histone methyltransferases (Fig. 5b). We also identified sequence specific DNA binding proteins, suggesting cohesive but varied mechanisms by which these genes may increase risk for 22q11.2DS-CTDs (Fig. 5a, b). Table 11. c Types of variants in chromatin genes (PTV is protein truncating, D-mis are damaging missense variants, mis is missense). A total of 57 PTVs were identified in 90 chromatin genes in sporadic CHD by the PCGC. In total, 14 PTVs were identified among 24 de novo variants in nine genes by Sifrim et al. A total of three PTVs were found among 42 variants in 22q11.2DS-CTDs. P-values derive from two-Proportions Z-Test. A total of 22 of the chromatin regulatory genes and five of the six recurrent genes were among the most Constrained Genes gene set (Fig. 5a), and this overlap was significant (overlap P = 1.27 × 10 −17 and 6.27 × 10 −6 , respectively). Additionally, nine are implicated in causing sporadic CHD from other studies in the literature 21,22,32 as shown in Fig. 5a (overlap P = 7.05 × 10 −5 ). Further, we show the overlap between the genes for sporadic CHD from the PCGC and the four from Sifrim et al. (Fig. 5a; Supplementary Table 11). We next examined expression levels of the chromatin genes found in 22q11.2DS-CTDs in cardiac progenitor tissues. TBX1 is expressed in the progenitor cells of the pharyngeal apparatus (PA) that migrate to the cardiac OFT during early embryogenesis 2 . A logical hypothesis based upon results of this study is that altered chromatin regulatory genes in 22q11.2DS-CTDs may modify expression of TBX1 and/or downstream genes, disrupting normal cardiac development. If this is the case, such chromatin modifiers should be expressed in the same cell types as TBX1 during the same developmental period. Most chromatin regulatory enzymes are widely expressed but they may show enrichment in certain cell types or tissues. We therefore examined expression in the PA, OFT + RV and LV + atria. We found 36 of the 39 genes show . The types of interaction evidence for the network edges are indicated by the line color (text mining, experiments, database, co-expression, neighborhood, gene fusion and co-occurrence). The high confidence score of 0.700 was used to create the network. Kmeans clustering was used to generate three clusters (cluster 1, red; cluster 2, blue; cluster 3, green). The nodes based on confidence; with line thickness indicates the strength of the support of the data in STRING. Six genes found by ORA are indicated (Italic blue font for gene names), chromatin genes found with de novo mutations in one or more cases with sporadic CHD (with asterisk next to the gene names). Constrained genes are indicated (Black circle surrounding nodes) and candidate sporadic CHD genes are shown (underscore for the gene names). b Representative most significant gene ontology terms from the STRING image, color coordinated according to the STRING image (molecular function, MF; cellular component, CC; biological process, BP). FDR, false discovery rate, -log10 P-value. enriched expression in TBX1 relevant tissues, specifically the PA as compared to either the OFT + RV or the LV + atria, where TBX1 is not expressed (Fig. 6; Supplementary Table 5) providing support for our hypothesis that chromatin regulatory genes might act in the genetic and epigenetic pathways of TBX1 function. DISCUSSION This is the largest study to date in which WGS was used to identify genetic modifiers of 22q11.2DS-CTDs. From this study, we uncovered most damaging rare variants (MDRVs) in chromatin regulatory genes that we suggest serve as modifiers in 8.5% of individuals with 22q11.2DS-CTDs. Some of the chromatin regulatory genes identified are co-expressed in the same cell type with TBX1, encoding a transcription factor and the main gene for 22q11.2DS. We generated a model to explain the shared types of chromatin regulatory genes and possible mechanisms by which some of these genes may interact with TBX1 (Fig. 7). We identified two main classes of chromatin regulatory proteins with MDRVs as illustrated in Fig. 7. Fig. 7a shows the histone acetyltransferases (HATs) that harbored MDRVs in this study. These HATs all share a MYST domain and have important roles in transcriptional activation 33 . We identified MDRVs in genes in two MYST different histone acetyltransferase families, the MOZ and NuA4 complexes (Fig. 7a). One of the genes in the MOZ complex is KAT6A. Of interest, it was found that Kat6a genetically interacts with Tbx1 in formation of the aortic arch in mouse models and the mechanism is by regulation H3K9 acetylation in the Tbx1 locus resulting in activation of gene expression 34 (Fig. 7a). It is therefore possible that MDRVs in these HATs mediate transcriptional activation of TBX1 and/or downstream genes. In Fig. 7b, we show the second main class of chromatin regulatory genes. We included lysine methyltransferases that had MDRVs and a possible role with TBX1 protein in regulating transcription of downstream genes. KMT2C and KMT2D are two methyltransferases that had recurrent MDRVs (Fig. 7b). Heterozygous gene inactivating mutations in KMT2D cause Kabuki syndrome in humans. KMT2C and KMT2D are part of COMPASS-Like family of H3K4 histone lysine methyltransferases 35 . It was shown that TBX1 physically interacts with KMT2C to mediate H3K4me1 methylation to activate downstream genes such as WNT5A 36 (Fig. 7b). The H3k4me1 mark is for poised enhancers that are further modified to promote gene expression 37 . Other proteins with MDRVs are members of the PRC (Polycomb) complex that represses transcription, the BAF complex that alters nucleosomes (SWI/SNF) 38 and NSD1 and NSD2, which are involved in lysine methylation 39 . EZH2 is a member of the polycomb complex that methylates H3K27 and Ezh2 genetically interacts with Tbx1 for mouse cardiovascular development 40 , supporting a connection to Tbx1. CHD7 is the gene for CHARGE syndrome and functions as an ATP-dependent chromatin remodeler that interacts with the BAF complex 38 as shown in Fig. 7b. TBX1 genetically and physically interacts with CHD7 for formation of the aortic arch 41 . Of interest, KAT6A and CHD7 were also among the 26 genes identified for tetralogy of Fallot patients based on reanalysis of whole exome sequence data from 811 subjects 42 . Therefore, some of the genes we identified have been shown to be in the molecular pathway of TBX1, supporting their candidacy as modifiers. De novo gene inactivating mutations in chromatin regulatory genes occurred in 3% of individuals in previous sporadic CHD studies from the PCGC 21,22,31 . Interestingly, we found an even greater occurrence of variants in chromatin regulatory genes in our 22q11.2DS cohort, with MDRVs identified in 8.5% of the 22q11.2DS-CTD cases. We previously identified chromatin regulatory genes from whole exome sequence from 89 individuals with 22q11.2DS and tetralogy of Fallot versus controls, although there were differences in the genes themselves 43 . Half of the genes that overlapped between our study, reported here, of 22q11.2DS-CTDs and studies of sporadic CHD from the PCGC, are causative of known syndromes in which CHD is an associated feature [25][26][27][28][29][30] 7 . Our study of 22q11.2DS-CTDs therefore provides more support for some of the chromatin regulatory genes and their contribution to the etiology of CHD. The results in this study and previous work 12,43 , supports the idea that genetic modifiers for 22q11.2DS-CTDs are inherited as a complex trait, similar to risk factors for sporadic CHD. However, in our study most of the variants were missense changes in chromatin regulatory genes that are predicted to be damaging, while for sporadic CHD, most were protein truncating gene inactivating mutations 7,22,32 . In addition to de novo heterozygous mutations in chromatin regulatory genes, recessive inherited cilia genes were identified in the PCGC study by taking gene set approaches 7,22 . Some of the subjects with recessive mutations in cilia genes had laterality defects such as heterotaxy, while others had CTDs 7 . We did not find evidence for heterozygous MDRVs in cilia genes using the same cilia gene sets as previously analyzed 7 and this suggests that 22q11.2DS-CTD modifiers do not parallel all the different types of genetic risk factors for sporadic CHD. One of the limitations of this study is that we have a relatively small sample size. In addition to our unbiased evaluation of genes with MDRVs that occurred recurrently in 22q11.2DS cases, we also tested the aggregated effect MDRVs at the gene set level. The gene set approach we used was similar to other study designs used to identify genes for sporadic CHD or specific subtypes of cardiac malformations (Supplementary Table 12). These gene set approaches analyze rare or ultra-rare coding variants. Similar functional annotation software and algorithms were used in our study of 22q11.2DS. The only difference was in our analysis of 22q11.2DS, we used newer machine learning algorithms to identify splicing variants. Some of the other CHD studies also used STRING software to help identify the functional characteristics of genes identified and several compared their findings to those of the Pediatric Cardiac Genomics Consortium or other studies of CHD 7,21,22,32 . We used gene expression in mouse embryos to prioritize genes identified for 22q11.2DS, based in part based upon the success of these studies that used a similar approach 21,44 . Other studies used human fetal heart gene expression in a similar way to our study that used mouse embryo gene expression. There was one study of a different genomic disorder, Williams-Beuren syndrome, to identify modifiers of supravalvular aortic stenosis (SVAS) 45 . This study of Williams-Beuren syndrome, examined rare coding variants using WES data from 58 cases with SVAS (16 had severe SVAS) and 46 controls, but was underpowered for statistical correction of all gene sets compiled by GSEA (Gene Set Enrichment Analysis). Nevertheless, the authors of the SVAS study identified gene pathways that might contribute to this phenotype 45 . We also note two studies in which a similar gene set approach was successfully undertaken of rare, potentially damaging variants to identify genes for schizophrenia 46 . Therefore, our study identified modifiers of CHD in 22q11.2DS using similar approaches to these reported in the literature with comparable success. By examining 19 gene sets, we found significant enrichment of MDRVs burden in chromatin regulating genes, which suggests it is not the overall burden but rather the regional MDRV burden of particular genes which plays critical role in TBX1 regulatory network as supported by the transcriptomic profiling data, that makes the cardiac phenotypic difference in the 22q11.2DS cohorts. The 19 gene sets in this study not only served as the testing unit to increase statistical power for the aggregated effect of mutations but also as gene level annotation, by which we found the chromatin regulatory genes identified for 22q11.2DS-CTDs also had appreciable constrained features, implicating their intolerance to variation. Furthermore, we didn't find enrichment among the gene sets in the 22q11.2DS control group. In summary, our findings suggest that disturbance of a TBX1 gene network by the presence of MDRVs in chromatin regulatory genes may disrupt cardiac development in a significant subset of individuals with 22q11.2DS, thereby serving as genetic modifiers. Since some of these chromatin genes were found in only one or a few individuals with sporadic CHD, our work provides more support for their candidacy as disease genes beyond many of their known roles in established genetic syndromes. This study therefore emphasizes the shared mechanisms involving the TBX1 gene network in CHD. Study cohort and definition of case vs control groups A total of 1595 subjects with 22q11.2DS were recruited by the International 22q11.2 Brain and Behavior Consortium (IBBC); details on ascertainment and original study design were provided previously [10][11][12][13] . The IBBC was a retrospective study of 1,595 individuals with 22q11.2DS in which WGS was performed to identify genes for schizophrenia [10][11][12][13] . We adopted the same definitions of CHD and CTD cases subset within the CHD group, and controls as described in detail in our previous study 10 and are provided in Supplementary Table 1. As described in ref. 10 , a subject with an intracardiac and/or aortic arch defect or defect in arterial branching from the aortic arch, were considered as a CHD case. A subject with a CTD excluded individuals with an isolated VSD or ASD. Given high prevalence in the general population 47,48 , individuals with common, minor heart anomalies (VSD that closed spontaneously in infancy, ASD that closed spontaneously in infancy, persistent foramen ovale that closed spontaneously in infancy, or bicuspid aortic valve, in the absence of other malformations), were considered to be controls. Of the 1595 22q11.2DS samples sequenced, 1182 passed the quality control (QC) procedures for the current study ( Fig. 1a; Supplementary Fig. 1). Briefly, samples with no CHD phenotype information, and those with a 22q11.2 deletion other than the typical 3 million base pair LCR22A-LCR22D deletion were removed. Contaminated and mixed samples as determined by Identity-bystatus based on common and independent variants were removed. Only one of related sample pairs and duplicated sample pairs was retained for analysis. After sample QC measures, the cohort consisted of 537 controls, and 645 CHD cases among which 456 have a CTD (Supplementary Table 1). Fig. 5a-STRING image) surrounding a representative nucleosome. a MYST family proteins involved in histone acetylation with respect to regulation of TBX1 expression. b TBX1 protein regulates expression of downstream genes including WNT5a, via chromatin regulators that belong to several classes as indicated. This is mediated in part, by physical interaction with CHD7. DNA is shown as a gray double helix. Gene activation is shown as an arrow and repression as a cross bar. Ethics, consent and permissions samples retrospectively collected with the written and signed informed consent of each of the subjects. Quality control of variants in raw WGS data Details on the variant calling processes were provided previously 10 . We performed a comprehensive QC analysis of the raw genotype data as described in Supplementary Fig. 1. First, variants in the low copy repeat (LCR) regions and 3 million base pair 22q11.2 deletion region were removed. Single nucleotide variants (SNVs) were removed if genotype rate was < 0.95 and Hardy-Weinberg equilibrium (HWE) < 10 −6 . Indels were removed if genotype rate was < 0.97 and HWE < 10 −5 . Of note, we adopted a more stringent filtering criteria for indels as compared to SNVs because variant calling pipelines usually have a slightly lower confidence in identifying indels accurately. Monomorphic variants in the remaining 1182 samples were then removed. A total of 21,695,115 of the 30,834,871 diploid variants in the raw data set passed QC procedures. The detailed composition of variants in the cleaned data are shown in Supplementary Fig. 2a. A total of 1,861,125 variants were identified as indels accounting for 8.62% of the total variation. Among them, 98.8% of the indels were within 10 bp, which most variant calling pipelines have good sensitivity and specificity to make a call accurately ( Supplementary Fig. 2b). Most of the variants were rare with an alternative allele frequency (AAF) < = 0.01 ( Supplementary Fig. 2c). Principal component analysis (PCA) Details about PCA for this study have been described previously in ref. 10 . Briefly, genome-wide diploid variants and the International HapMap project phase III release 3 data were used for PCA. Shared variants in our dataset and the HapMap dataset were extracted and combined into one dataset. Variants with A > T, T > A, G > C and C > G allele types were removed to avoid DNA strand-flip problems. Variants that had a minor allele frequency of <0.05 and all variants on chromosome X and Y, were excluded. The autosomal common variants were filtered using the -indep function of PLINK to ensure only independent variants were used for PCA. Then PCA was conducted with the -pca function of PLINK1.9 beta. A total of 879 (74.4%) subjects were Caucasian, 184 (15.6%) were of African descent or admixed and 119 (10.0%) were Hispanic ( Supplementary Fig. 3a). Further PCA was stratified by case and control status ( Supplementary Fig. 3b) demonstrating that cases and controls are well matched by ancestry, indicating low probability of population stratification. Variant annotation and identification of the MDRVs Protein truncating variant (PTVs) i.e., Loss of function (LoF) variants 49 including indel-frameshift, stop gain, splice donor, splice acceptor, stop loss, start loss were annotated in parallel by Variant Effect Predictor (VEP) plugin LOFTEE 50 , Bystro 51 and ANNOVAR software 52 and LoF variants were combined as shown in Supplementary Fig. 4 . Damaging missense variants were annotated by ensemble score of MetaSVM 53 . Damaging splicing variants were annotated by spliceAI 54 and two ensemble scores, ada 55 and random forest (rf) scores from the Ensemble database available from dbscSNV 56,57 . We are aware that the annotated functional variants may be enriched for false positives, for example, of sequence artifacts. Therefore, the putative functional variants were subjected to further filtering-based annotation to remove potential false positives and to obtain high-quality MDRVs. First, relatively less conserved variants were excluded at a phastCons score < 0.5 and putatively less deleterious variants were removed with CADD score 58 < 10. Secondly constrained coding regions (CCRS) 59 restriction was further applied to prioritize and refine the MDRVs. In silico validation of the variants using GATK Variants were called by GATK from the alignment to a Human Reference Genome (GRCh37/hg19) following default parameter settings in GATK release 4.0.0. To cross-compare variants from the two different pipelines, variants from PEMapper and PECaller were first lifted over from hg38 to hg19 (https://genome.ucsc.edu/cgibin/hgLiftOver). The methods used to cross compare the two calling variants are shown in Supplementary Fig. 5. A small proportion, 0.15% comprising 32,291 of 21,695,115 QC'd variants failed the liftover to hg19. The VCF files were split into individual VCF files for each sample, respectively. A total of 1151 samples had variants/VCF files from both pipelines. Next, each pair of VCF files were cross-compared using Bcftools (https://samtools.github.io/ bcftools/bcftools.html) as evaluated by concordance of chromosome, position, reference allele, alternate alleles and genotype i.e. in genotype mode or just the first four parameters (genotype mode =0). Indels were deemed as validated if > = 10% of the base pairs overlapped. Gene level annotation RNA-seq analysis of mouse embryonic cardiac development tissues: Genes contributing to CHD should be expressed in the embryonic pharyngeal apparatus, aortic arch and/or developing heart 21 . We performed RNA-seq analysis using total RNA from the micro-dissected tissues of mouse embryos. This study is part of the Albert Einstein College of Medicine, Institutional Animal Care and Use Committee, 00001034. Wild type mouse embryos in a mixed SwissWebster/C57Bl/6 background at E10.5 (30-32 somite) were isolated in ice-cold Dulbecco's phosphate-buffered saline (DPBS), then the pharyngeal arches 2-6 (PA2-6; cut along dorsal aorta), outflow tract together with the right ventricle (OFT + RV), and left ventricle plus atria (LV+Atria) of embryos were immediately micro-dissected and then were frozen separately and stored in the −80°C freezer. Each sample (each biological replicate) for RNA isolation was pooled from four embryos and three biological replicates were prepared for the RNA-seq. Total RNA was extracted using TRIzol and miRNeasy Mini Kits (QIAGEN, 217004), and oncolumn DNase I digestion (QIAGEN, 79254) was performed before eluting RNA in the RNase/DNase free water and all samples passed quality control measures. Library preparations and Illumina sequencing were performed at the Einstein Epigenomics Core Facility (https://einsteinmed.org/departments/genetics/resources/ epigenomics-core.aspx). The libraries were prepared using the KAPA stranded RNA-seq Kit with RiboErase (HMR) (KAPA-Roche) following the protocol provided in the Kit. Ribosomal RNA was removed before library preparation. Quality of libraries was examined by Qbit (fluorometric quantitation; Invitrogen), bioanalyzer (Agilent 2100) and qPCR (Roche light cycler), and all passed quality control. Then the libraries were multiplexed and sequenced using the Illumina NextSeq 500 system as 2 × 150 bp paired ends. For RNA-seq analysis, the mean gene expression level from the three tissues was used to prioritize protein-coding genes, and the log transformations of the mean expression levels were used as weights for the weighted gene set based test. Gene constrained scores: Genes that are crucial for the development of an organism and survival will be depleted of LoF variants in natural populations, whereas non-essential genes will tolerate their accumulation 50 . We used the latest gene-level constraint score, rank summation with re-sorting (VIRLOF) 60 , which is based on deeply curated LoF variants from 125,748 exomes and 15,708 genomes of sequence data aggregated by gnomAD and the combination of LoF observed/expected upper bound fractions (LOEUF) 50 and gene variation intolerance rank (GeVIR) 60 . Gene sets (19 sets in three categories) Three constrained score-based gene sets: The Constrained Genes gene set was curated from the top 5th percentile of VIRLOF gene level matrix (n = 968). These data were derived from population genetics approaches 50 . The Essential Genes gene set included the shared core set of essential genes derived from three independent large-scale screens to assess the effect of single-gene mutations on cell viability or survival of haploid human cancer cell lines, cell-based essentiality 61-63 (n = 956). The Haploinsufficiency Genes gene set was compiled based on the top 5th percentile (n = 767) from the genome-wide haploinsufficiency score (HIS) 19 . Two chromatin related gene sets: It has been established that de novo variants in chromatin related genes are associated with the pathogenesis of sporadic 21,22 and syndromic CHD 23 . We therefore downloaded all genes with chromatin function and modification related terms from GSEA (https://www.gseamsigdb.org/gsea/index.jsp), totaling 866 chromatin related genes from 81 GO and REACTOME terms including chromatin (de) acetylation, (de)methylation, (de)phosphorylation, (de)ubiquitylation, chromatin remodeling, chromatin assembly or disassembly. We named this gene set as All Chromatin Regulatory Genes (n = 866). We note that the small Chromatin Genes (n = 163) gene set used by the PCGC comprises a small subset of the All Chromatin Regulatory Genes gene set. For the second gene set, we focused on one Reactome term, Chromatin Modifying Enzymes (n = 274), because it included many chromatin modifying enzymes. The first and larger gene set covers most chromatin genes and serves as a hypothesis-free gene set, while the second, represents a smaller subset of the first. Eleven gene sets with recognized relevance to pathogenesis of sporadic CHD and three non-cardiac gene sets: From an original 19 gene sets previously curated and employed to investigate the genetic architecture of sporadic CHD 7 , we selected 14, each having over 100 genes. These include 11 known pathways with recognized relevance to the pathogenesis of sporadic CHD, and three non-cardiac control gene sets. The cardiac gene sets include: Candidate CHD Genes (excluding cilia genes, n = 402), Chromatin Genes (n = 163), Cilia Genes (n = 669), SysCilia Genes (n = 302), which is a subset of Cilia Genes, Notch Associated Genes (n = 130), TGF-B Genes (n = 431), Cytoskeletal Genes (excluding cilia, n = 791), WNT Signaling Genes (n = 297), Hedgehog Signaling Genes (n = 149), FoxJ1 Genes that consist of mobile cilia genes (n = 116), Platelet Derived Growth Factor (PDGF) Signaling Genes (n = 116). The three non-cardiac gene sets are Axon Guidance Genes (n = 191), Brain Expressed Genes (n = 400) and Liver Expressed Genes (n = 400), serving as controls. Mouse gene liftover to human orthologs: Mouse genes were lifted over to human orthologs using the R package, biomart 11,64 . For genes with either multiple mapping or missing mapping, web crawler was applied to automatically search for possible orthologs in NCBI genes (https://www.ncbi.nlm.nih.gov/gene). Over-representation Analysis (ORA): ORA is a widely used approach to determine whether known pathways or gene sets are over-represented (enriched) in an experimentally derived gene list 65 . We adopted a stringent criterion to first select only genes identified to have one or more MDRV in two or more CTD cases and no MDRV in controls (n = 0 in controls). A similar criterion was used to select recurrently affected genes in controls, i.e., genes identified to have one or more MDRV in two or more controls and no MDRV in CTD cases (n = 0 in cases). As most of the affected genes have one identified MDRV, these criteria can account for potentially random background noise for false positive MDRVs in the whole 22q11.2DS cohort. These restricted subsets of recurrently affected genes in the CTD group, and in the control group, were then evaluated by ORA for enrichment in the 19 gene sets independently. The ORA can be calculated by hypergeometric distribution 66 : where N is the total number of genes included in the analysis as the background distribution; M is the final number of genes for each gene set among N, n is the number of recurrently affected genes; and k is the overlap of n and M, which is among the recurrently affected genes. For example, in the scenario of expressed genes in cardiac progenitor cells, in the CTD case group, N = 413, M = 27 for Chromatin Regulatory Enzymes (n = 274), n = 12 and k = 6; while in the control group, N = 466, M = 30 for Chromatin Regulatory Enzymes (n = 274), n = 13 and k = 1. Weighted gene-based test: To reduce the signal/noise ratio, we employed a weighted gene-set based test to examine the aggregated burden of MDRVs in CTD cases as compared with controls. Specifically, let Z be k 1 vector of z statistics of genes in a gene set obtained from the gene-based test for each gene, and w be k 1 weighting vector of corresponding importance scores, more specifically, gene expression level in developing heart and related tissues that provide clues. Let J = ww T w T w be a k k importance score matrix. The statistic is defined as T ¼ Z T JZ. For the 1861 high-confidence MDRVs, we first performed a gene-based Fisher's Exact Test (FET) between CTD cases and controls to obtain the gene-level P-value and OR solely based on genetic data, then transformed this to the individual gene level statistic z. The log transformed mean expression level as described above was used as w, i.e., a given gene was assigned higher weight if it had a higher expression level in progenitor cells for cardiac development from RNA-seq analysis of E10.5 mouse embryos. An empirical gene set P value was calculated using 2000 permutations by randomly shuffling case and control status. In each replicate, a T statistic was generated as described above for each gene set. The empirical P value for each gene set was calculated as P value = n/N, where n is the number of test-statistics as or more extreme than the observed test statistic and N is the total number of random permutations. Multiple testing correction: We adopted the false discovery rate (FDR) to control for multiple testing burden. For the weighted gene set based test, we corrected for 19 total P values. For ORA, we adjusted for 19 4 = 76 total P values for the CTD and control group (×2), for filtered genes based upon expression and all genes (×2). Reporting summary Further information on research design is available in the Nature Research Reporting Summary linked to this article. DATA AVAILABILITY Data supporting the findings of this work are provided in Supplementary Data Tables. The RNA-seq data is available at the NCBI (National Center for Biotechnology Information) Sequence Read Archive (SRA) number: PRJNA885469. The raw WGS data for the 22q11.2DS cohort was previously deposited on NIMH Data Archive (https:// nda.nih.gov/study.html?id=938). Data access requires Institutional approval and requests will be reviewed by the NIMH Data Archive Data Access Committee. CODE AVAILABILITY The core computational pipelines used for quality control, annotation, and statistical analysis of the WGS data was coded in python 2.7. The weighted gene-set based analysis and plot of all figures were done in RStudio. Both the python code and R code are available at https://github.com/bioinfoDZ/22q11wgs.	2023-07-19T13:06:13.045Z	2023-07-18T00:00:00.000	{ "year": 2023, "sha1": "ba71361945117de7e4a7de54af0f54f9bc2e5ba0", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41525-023-00363-y.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "3ef205dabf96e35301ae711ab04236a97d2752a7", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
14103582	pes2o/s2orc	v3-fos-license	A childhood case of symptomatic essential and psychogenic palatal tremor Palatal tremor is a rare movement disorder characterized by rhythmic contractions of the soft palate. It is most often symptomatic, secondary to brainstem or cerebellar disease and, in rarer cases, is categorized as essential in the absence of documented brain lesions. There have also been reports in the literature of cases of palatal tremor described as psychogenic because they were associated with psychological or psychiatric disorders. We describe the case of a 12-year-old boy with palatal tremor presenting clinical features of symptomatic essential and psychogenic palatal tremor, thus suggesting a neuropsychopathological continuum between the different forms of disease. Introduction Palatal tremor is a rare movement disorder characterized by rhythmic contraction of the soft palate. Palatal tremor is most often symptomatic, secondary to brainstem or cerebellar disease and, in rarer cases, is categorized as essential in the absence of documented brain lesions. 1 Some authors have documented hypertrophic degeneration of the inferior olives in secondary palatal tremor, but its precise role in causing palatal tremor has not yet been clarified and this finding is currently controversial. 2 Essential palatal tremor generally affects children of both genders, whereas secondary palatal tremor is most commonly observed in adult males. Essential palatal tremor is bilateral and usually disappears during sleep, whereas secondary palatal tremor is more frequently unilateral and persists even during sleep. Patients with essential palatal tremor usually have an ear click, which is absent in the symptomatic form. Although essential palatal tremor has a benign course and usually disappears spontaneously, it is very annoying for the patient, so several treatments have been administered, including anticonvulsants and anxiolytics, with varying therapeutic effects. At present no specific treatment has been established, although successful treatment by botulinum has recently been reported. 3 There have also been reports in the literature of cases of palatal tremor which have been described as psychogenic because they were associated with psychological or psychiatric disorders. In this form, known as psychogenic palatal tremor, the abnormal movement is intermittent, increases during attention, decreases during distraction, and is under voluntary control. 4 We describe the case of a 12-year-old boy with palatal tremor presenting clinical features of symptomatic essential and psychogenic Case report The case was a 12-year-old boy, the only child of unrelated parents. There was a positive family history of psychiatric disorders, in that the father has a personality with borderline and narcissistic traits, poor impulse control, and multiple motor tics, and the mother had a major, recurrent depressive disorder, and histrionic and paranoid personality traits. A maternal aunt suffered from a severe bipolar disorder. Moreover, the emotional climate in this family was characterized by a high level of "expressed emotion" with over involvement and criticism, and with a very conflictual relationship between the parents. 5 The patient was born by elective cesarean delivery at the eighth month of gestation after a pregnancy characterized by gestational diabetes treated with insulin. He showed mild motor developmental delay and inadequate performance at primary school, and difficulty in relating to peers. He came to our attention about one year earlier due to the emergence of behavioral disturbances characterized by oppositional defiant behavior and aggression, episodes of psychomotor agitation with dysphoria, and aggression towards others. About two years before these events, when the patient was 10 years old, his parents had started to hear a clicking sound coming from the boy, but no research was carried out until our observations. No drug therapy had been taken. We performed clinical and instrumental assessments, including physical, neurological, and otolaryngological examinations, laboratory tests, audiological and neuroimaging studies, and a psychodiagnostic assessment, including the IQ rating scale (Wechsler Intelligence Scale for Children III) and interviews administered to the patient and his parents using the Child Behavior Checklist and Schedule for Affective Disorders and Schizophrenia for School-Age Children -Present and Lifetime Version. The neurological examination showed only hypotonia of the facial muscles causing drooling, but no other clinical signs, except for the presence of rhythmical bilateral palatal movements, mainly on the right side, associated with a clicking sound that could easily be heard when listening close to the child's ear. No ataxia or nystagmus was present. The clicking was continuous and audible by his parents even during sleep. The patient seemed to be very annoyed and concerned about this ear click, which was present with the mouth both open and closed. Nevertheless, it was under voluntary control, and the patient was able to voluntarily induce it or stop it for a few minutes. Moreover, the ear click was less audible or had a changed frequency during the performance of complex motor and cognitive tasks. Otolaryngological examination with fibroscopy detected a bilateral palatal tremor at a frequency of 80 cycles per minute. Laboratory tests, including a complete blood count, antinuclear antibody studies, thyroid tests, plasma copper, ceruloplasmin, plasma amino acids, serum lactate and pyruvate, ammonium, and transaminases, were normal. Audiometry and electroencephalography were also normal, as were brain magnetic resonance imaging, angiographic magnetic resonance imaging, and proton magnetic resonance spectroscopy, focusing particularly on study of the brainstem. During spontaneous observation, the patient showed a reduced adaptive and relational capacity, with his relationships being characterized by oppositional behavior that sometimes culminated in episodes of aggression and psychomotor agitation. His moods and affectivity were extremely unstable, featuring sudden swings and a poor control of emotions and impulses, in particular at moments of frustration and deferral of pleasure. Psychosomatic and anxiety symptoms were also detected. At the WISC-III test, he obtained an IQ of 65 (verbal IQ = 81, performance IQ = 56), expressing mild mental retardation. During the previous year, several therapies had been attempted, including anticonvulsants and antipsychotic drugs, with variable results. Topiramate and valproic acid treatment showed no effect on either the behavioral disturbances or the palatal tremor, and carbamazepine therapy showed a transient effect that disappeared for about two months but then recurred. Due to an increased frequency of episodes of agitation, chlorpromazine therapy was started, then carbolithium was added, resulting in an improvement of the behavioral disturbances. Treatment with piracetam, the effectiveness of which in essential palatal tremor was recently reported by Campistol-Plana et al, 3 was ineffective. It was not possible to administer treatment with botulinum toxin due to the patient's limited cooperation. All drug treatments were associated with psychotherapy. After six months of treatment with carbolithium, chlorpromazine, and family psychotherapy, the behavioral disturbances were further improved and the ear click was audible only on the right side and had started to fluctuate, disappearing spontaneously for days, reappearing spontaneously in stressful situations. Over time, the patient had developed a greater ability to inhibit the palatal tremor voluntarily and to induce it on request Dovepress Dovepress 225 Palatal tremor or by thinking about it, sometimes to attract attention, thus demonstrating voluntary control. Discussion Palatal tremor is an abnormal movement of the soft palate and was previously known as palatal myoclonus but more recently has been renamed palatal tremor because this corresponds better with the electrophysiological characteristics of the rhythmic pattern. The heterogeneous nature of palatal tremor covers a wide spectrum of clinical pictures, including symptomatic essential and psychogenic forms. Secondary palatal tremor can be a consequence of vascular, infectious, degenerative, traumatic, or neoplastic lesions of the brainstem or cerebellum. This form is often associated with neurological deficits and, once present, persists throughout life and does not disappear during sleep or even in coma. Most symptomatic cases occur in adulthood, while a few cases have been reported in childhood in the course of Krabbe's disease, 6 a cerebellar tumor, 7 and encephalitis. 8 In secondary palatal tremor, abnormal hypertrophy of the inferior olives has been detected, but its precise role in causing palatal tremor is controversial and not yet clearly demonstrated. Recently Shaikh et al 9,10 have shown, through a mathematical model, that oculopalatal tremor oscillations originate in the hypertrophic inferior olive and are amplified by learning in the cerebellum. In essential palatal tremor, neurological investigations and brain neuroimaging are normal, while an ear click is present, perceived as objective tinnitus, that does not occur in the symptomatic form. Deuschl et al 11 demonstrated that, in secondary palatal tremor, the movement of the soft palate is caused by contraction of the levator veli palatini and in essential palatal tremor, by contraction of the tensor veli palatini which induces secondary closure of the Eustachian tubes, causing a clicking sound that is audible by the examiner. Essential palatal tremor is more frequently bilateral, occurs mainly in children, and usually disappears when the patient's mouth is open and during sleep, although in some cases it persists even during sleep. The pathophysiology of essential palatal tremor remains unclear; it is generally benign, persists for months or years, and in children it tends to fluctuate and disappear spontaneously. Essential palatal tremor has been observed in monozygotic male twins, in one of whom symptoms developed at the age of 14 years and in the other at 37 years, suggesting a genetic etiology of the disorder. 12 Fernandez-Alvarez 13 included essential palatal tremor in the class of transient movement disorders, a heterogeneous group of primary pediatric movement disorders with a limited duration of expression over time, none or mild disability, and a spontaneously fading disorder. The diagnosis is clinical in the absence of biological markers. Voluntary control of palatal tremor is possible in some cases of essential palatal tremor. In fact, the abnormal movement can be voluntarily induced or inhibited, and can also be discontinued when the patient is concentrating on motor and cognitive tasks, performing movements, reading, thinking, or doing calculations. Some authors have suggested that voluntary control of essential palatal tremor could be due to a certain degree of cortical control that is otherwise absent in the symptomatic forms, while other authors have suggested that it could be due to a psychogenic origin of the disorder. Few cases classified as psychogenic palatal tremor have been reported in the literature. This form is often associated with psychopathology, responds to placebo and improves with psychotherapy. Furthermore, the movements have a variable frequency, and unlike the other forms of palatal tremor, increase during attention and decrease when the patient is distracted. Nevertheless, psychiatric symptoms, in particular anxiety disorders, have also been reported in the literature for some cases of essential palatal tremor. After a review of the literature in relation to different forms of voluntary control, Samuel et al 14 suggested that rather, than being a uniform pathogenic entity, essential palatal tremor may be a heterogeneous disorder including different forms; in particular, a form of conversion disorder, a tic disorder, or a series of completely involuntary movements, as described in most patients with essential palatal tremor. Moreover, the same authors suggested a correlation between secondary palatal tremor and essential palatal tremor, in view of the fact that functional magnetic resonance imaging studies have shown an activation of the inferior olives during palatal movements in the essential form. 15 These studies suggest an involvement of the same structures in essential palatal tremor and secondary palatal tremor. In addition, in some cases of essential palatal tremor, an association with minor ailments has been reported, including respiratory infections, otitis, tonsillitis, fever, minor head trauma, and headache. The clinical characteristics of our patient seem to support this hypothesis of a continuum between the different forms of palatal tremor (Table 1). He has a palatal tremor with an ear click, audible in both ears, which is usually submit your manuscript \| www.dovepress.com 226 Margari et al present in the essential form but is rare in the symptomatic form. The palatal tremor persisted during sleep, which is common in the symptomatic form but rare in the essential form. Furthermore, neuroimaging studies did not document lesions in the brainstem or cerebellum, which are typically present in the symptomatic form. In addition, in our patient, pre-perinatal risk factors were present, namely a family history of psychiatric disorders, motor developmental delay, mental retardation, hypotonia of the facial muscles, and behavioral disturbances. These manifestations suggest symptomatic involvement of the brain. However, our patient was able to voluntarily induce or stop the tremor. Even if voluntary modulation of palatal tremor has been reported in literature in some cases of essential palatal tremor, our patient has developed possible voluntary control, suggesting the acquisition of special motor skills rather than an inhibition of involuntary movements, and this seems suggestive of a psychogenic etiology. 16 Limited data are available about psychogenic movement disorders in children, but clinical findings suggesting a psychogenic cause similar to that in adults, including inconsistent character of the movements, exacerbation with stress and while paying attention, and disappearance when distracted, as well as a response to placebo and improvement with psychotherapy. The underlying psychiatric diagnosis was a conversion disorder in most of the cases described in literature. 17 In our patient, the disorder also shared some features of psychogenic movement disorders, such as a discontinuous ear click, association with psychopathological disorders, an increase during periods of attention and disappearance at moments of distraction, as well as voluntary control of the movement. In conclusion, the clinical characteristics of our patient suggest that there may be a neuropsychopathological continuum between different forms of palatal tremor, but this hypothesis requires further investigation, especially functional neuroimaging and genetic studies.	2017-04-04T16:28:23.539Z	2011-04-27T00:00:00.000	{ "year": 2011, "sha1": "d606e6ac53ffbbc01ffcfc614d76f7533f1eecb8", "oa_license": "CCBYNC", "oa_url": "https://www.dovepress.com/getfile.php?fileID=9855", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "5a3639ac583d307b09afadbbf15ea05b2d4b16be", "s2fieldsofstudy": [ "Medicine", "Psychology" ], "extfieldsofstudy": [ "Medicine" ] }
271906842	pes2o/s2orc	v3-fos-license	Extended High-frequency Audiometry in the Elderly: A Narrative Review Extended high-frequency audiometry (HFA) is considered an important tool in the detection of hearing loss. However, the values at extended high frequencies (EHF) in older adults (in both men and women) are associated with considerable uncertainty due to limited reference data. The presented review aimed to analyze hearing thresholds at EHF in adults older than 60 years. A literature search for HFA-related keyword combinations was conducted using the electronic databases PubMed, Scopus, and Web of Science. A total of 1654 records, published in the last 22 years, were identified through this search, of which only 7 articles were ultimately included in the analysis. Multiple studies have shown that significant hearing loss can be observed at EHF in older adults. Hearing thresholds in the frequency range of 9-20 kHz in the elderly varied widely across the studies. Therefore, further research in this field is needed to complete the normative data. INTRODUCTION The average life expectancy is increasing globally. 1Hand in hand with this increase, however, comes a rise in the incidence of hearing loss. 2 Based on the reports of the World Health Organization, nearly 2.5 billion people (i.e., 1 in 4 people) are projected to have hearing problems by 2050. 2 Currently, 430 million people live with disabling hearing loss globally. 3Approximately 30% of people over 60 years of age have hearing loss. 3Age-related hearing loss (ARHL, presbycusis) is the most common hearing disorder and a major cause of chronic disability in older age.Age-related hearing loss can cause difficulty in speech comprehension, thereby compromising communication skills and potentially leading to social isolation and loneliness. 4In the early stages, ARHL typically affects audibility of higher frequencies, but over time, the impairment spreads to medium and low frequencies. 4In urban settings, hearing impairment is very common, as intense community noise exacerbates the deterioration of hearing with old age. 4 examine hearing thresholds, pure-tone audiometry is widely used.Based on the examined frequency range, we can distinguish between conventional audiometry (CA, frequency range of 0.125-8 kHz) and extended high-frequency audiometry (HFA, frequency range of 9-20 kHz). 5][8][9][10] However, the term "extended high frequencies" (EHF) is typically used for the frequency range above 8 kHz nowadays. 11,12For the purposes of this review, we will follow this definition. The hearing thresholds at EHF are still affected by a great deal of uncertainty, especially for the older age groups. 13Currently, the literature on audiometry above 8 kHz, up to 20 kHz, is limited.Of the available works, most focus on young or middle-aged participants, i.e., adults under 60 years of age.Not many studies with participants over this age have been published yet, and, in effect, hearing thresholds at all EHF could not yet have been properly determined for them.The causes of the low number of such studies include the higher prevalence of moderate to severe hearing impairment in people over 65 years of age compared to younger adults 14 or otological diseases, including tinnitus, 15 as well as cognitive impairment often present in these individuals, 16 which makes recruitment of a suitable population more difficult. The International Organization for Standardization (ISO) provides in its Standard 7029:2017 information on the expected median values at audiometric frequencies of 9-12.5 kHz for age groups of 22-80 years. 13However, the values at this frequency range in the age category of 80+ years are associated with considerable uncertainty in both men and women due to the aforementioned low amount of available data.At the frequency of 12.5 kHz, the same is true for all age categories from 30 years onwards.No normative data are then available for any of the EHF over 12.5 kHz. 13is article aims to (i) summarize the knowledge on high-frequency hearing in the elderly and organize the information gained in this area over the last 22 years.Furthermore, we aim to (ii) provide readers with hearing threshold values at EHF for the oldest age categories given the fact that normative values are not fully established for that group yet, and, lastly, (iii) to draw attention to the importance of research in this area so that the mentioned standard thresholds can be properly determined and, subsequently, updated over time. Search Strategy A literature search was conducted using the electronic databases PubMed, Scopus, and Web of Science.Keyword combinations related to extended HFA in the elderly (see below) were searched. Inclusion Criteria Eligible literature was selected using the following criteria: (i) articles had to contain the keywords "EHFs, " "HFA, " "pure-tone audiometry, " and had to be concerned with assessing hearing thresholds at frequencies over 8 kHz in elderly (or seniors), using extended HFA; (ii) articles had to be published between 2002 and 2024; and (iii) only articles in English were included.(iv) Regarding the age of the participants, the studies had to be performed on study groups that included elderly adults with the age range of at least 60-68 years and older; (v) only original research articles were included; and (vi) finally, only studies where the authors examined at least 3 EHFs within the range of 9-20 kHz were included. Exclusion Criteria From records identified through database search, duplicates found in more than one of the databases were removed, as well as book chapters, conference proceedings, manuals, and brochures.Studies assessing the impact of hearing impairment on participants' quality of life, mental well-being, social contacts, or social life, as well as studies assessing the hearing of respondents in association with their acute or chronic diseases, injury, surgery, or treatment, were excluded.We did not include studies in which the authors primarily aimed to assess therapeutic and rehabilitation methods for restoring or improving participants' hearing or assess the hearing quality in relation to the use of hearing aids.Additionally, animal studies were excluded.Finally, the articles not clearly stating the examined frequencies and/or not adequately describing measurement conditions or participants were also excluded. RESULTS Using the above keywords and the 2002-2024 timeframe, a total of 1654 records were identified through database search.A total of 1620 records were excluded, as they were not directly related to the issue of interest.Subsequently, 9 duplicates found in more than one of the databases were removed, as were 18 articles not meeting the above inclusion criteria.Seven articles were ultimately included in the analysis and divided into 2 subgroups according to the age of participants.The entire search process is shown in Figure 1. Group 1 (n = 2) includes studies with participants older than 60 years only. 17,18Group 2 (n = 5) includes studies comprising, besides the elderly group over 60 years of age (minimum age range of 60-68 years), also younger participants. 5,11,12,19,20The most important characteristics of the identified studies classified into the groups above are summarized in Table 1. The studies were conducted in China (n = 2), 11,12 Japan (n = 1), 20 the United States (n = 1), 18 Spain (n = 1), 5 Norway (n = 1), 17 and the Czech Republic (n = 1). 19e number of participants in the identified studies ranged from 37 12 to 645. 5 The age of participants ranged from 5 years 5 to 90+ years.17 Regarding sex distribution, the proportion of men varied depending on the study type, ranging from 27% 12 to 57.4%.13 which provides descriptive statistics of the hearing threshold deviations for populations of otologically normal persons of various ages.The authors found that hearing thresholds at tested high frequencies worsened with increasing age, and the difference between hearing thresholds at EHF in otologically normal men and women was not significant. 17 A year later, the results of a longitudinal study aiming to analyze the hearing thresholds in 188 elderly individuals aged 60-81 years using CA and extended HFA (at frequencies of 9 kHz, 10 kHz, 11 kHz, 12 kHz, 14 kHz, 16 kHz, and 18 kHz) were published by Lee et al, who investigated longitudinal changes in hearing thresholds and the effects of age, sex, initial results, and past noise exposure.18 The authors found the rate of changes in hearing thresholds to be affected by age, sex, and initial threshold levels. O average, the hearing thresholds increased by approximately 1 dB per year in the elderly. 18lder women had a faster rate of changes at frequencies of 0.25-3 kHz (CA) and 10-11 kHz (extended HFA) compared to younger women (aged 60-69 years).Older men had a faster rate of changes at 6 kHz compared to younger men aged 60-69 years. Coparing men and women, a faster rate of changes at the higher frequencies of 6-12 kHz in women was found.No significant effect of past noise exposure on the rate of hearing threshold changes was observed by the authors. 18 Studies Including Participants From All Age Categories In 2011, Kurakata et al 20 published the results of their examination of hearing thresholds at frequencies ranging from 125 Hz to 16 kHz in Japanese adults.The study included 490 otologically normal participants, with 210 men and 280 women aged 18-89 years.Their results showed that the median hearing thresholds of Japanese participants were lower than the ISO values.They also pointed out that the responses of older participants could be unreliable at the highest frequencies. 202014 Czech study aimed to determine reference values of hearing thresholds at EHF above 8 kHz using CA and extended HFA (frequency range of 9-16 kHz) in 411 otologically normal individuals within the age range of 16-70 years.19 The authors compared the results with ISO 7029 13 and calculated the coefficients of quadratic, linear, polynomial, and power-law approximations for the measured frequencies.These presented results can be used to determine normal hearing thresholds at high frequencies for individual age groups.19 In the same year, Rodríguez et al 5 published the results of their prospective study with 645 otologically normal participants aged 5-90 years.The purpose of their study was to investigate and determine reference values of hearing thresholds at frequencies of 0.125-20 kHz, using CA (frequency range of 0.125-8 kHz) and extended HFA (frequency range of 9-20 kHz).Their study also aimed to compare the results with existing values at EHF. 5 They found that the measured hearing thresholds worsened with increasing frequencies and increasing age.5 In 2021, the Chinese study by Wang et al 11 Two years later, the study by Guo et al 12 was published. Th aim of this study slightly differed from previous studies. 12We have, nevertheless, decided to include it in our review because it met our criteria.The authors conducted a study involving 37 participants with an age range of 19-68 years with vestibular migraine (VM) and probable VM to determine their hearing thresholds at conventional and high frequencies, including 3 EHFs (10, 12.5, and 16 kHz) and confirmed the usefulness of extended HFA for early detection of hearing loss.The authors concluded that persons with VM tend to have bilateral hearing loss at EHF. 12 However, the number of participants older than 60 years was low, as well as their reported percentage of responses at EHF, 12 so we could not include their results in the analysis below. Analysis of Hearing Thresholds in the Selected Studies The hearing thresholds of the elderly varied across the included studies.This may be caused by the different composition of the study groups, the different numbers of participants, and possibly socio-cultural differences.For the purpose of our review, we have categorized the participants of the included studies by sex and by age into 2 groups, i.e., younger elderly (60-70 years) and older elderly (70+ years).If the authors of the included studies did not analyze the hearing thresholds of individual subcategories, we reported "all participants" (see Table 2 and Figure 2).Almost all studies, with one exception, reported hearing thresholds for the left and right ear combined.In one study, authors reported hearing thresholds in otologically normal (n = 60) and all participants for the left and right ear separately. 17We have decided to present the results of this one study as hearing thresholds of the better hearing ear (BHE) in otologically normal participants (Table 2 and Figure 2).An overview of hearing thresholds across the included studies in relation to age and sex is shown in Table 2. Figure 2 below illustrates the distribution of hearing thresholds values reported in the literature. DISCUSSION Extended HFA can detect hearing loss before it starts to affect low or middle frequencies, where the human voice frequency range spreads out.Some authors considered the frequency range of 4-8 kHz or 6-8 kHz as high frequencies, [6][7][8][9][10] while others designated the frequency range of 9-20 kHz as high frequencies, often defined as "extended high frequencies" (EHF).Only studies using the latter definition were included in our review. In the elderly, many age-related changes are observed in the organism, with changes in hearing thresholds being one of the prominent ones.So far, there are no internationally determined standard values for most EHF in this age group. 13Not many authors have measured HFA in the elderly over 60 years of age.Only 7 such studies were identified from the last 22 years and included in our review.The reasons can be easily assumed-at such an old age, there is a very small chance that the respondents would reliably hear high-frequency tones.However, if these measurements are not performed and published, normative values can never be established.For this reason, we appeal for the usefulness of extended HFAs even for those over 60 years of age to inform future meta-analyses and the subsequent establishment of normative hearing thresholds for these individuals.Our review shows that younger elderly (under 70 years of age) are able to hear surprisingly high-pitched tones, and although the results are poorer in the older age group (over 70), good hearing is not impossible even in that age group. The results vary widely across studies (Figure 2).Admittedly, most of them did not aim to focus primarily on hearing in the elderly; only 2 did so. 17,18Different sample sizes may account for the differences in results.It is evident that the authors certainly tried to make their study groups as large as possible, but in the elderly, the possibilities to perform an audiometric test are limited, and the conditions more difficult because of, e.g., presbycusis, possible cognitive impairment, and/or poor compliance (as mentioned above). Extended HFA is a very topical issue, so there should be no hesitation in obtaining data that could enable establishing norms for all frequencies and all age categories without limitation. Limitations The fact that only 7 studies were ultimately included in the review can be considered a limitation by some.However, we consider this rather a strength of our review as we can be certain that it only contains relevant studies.Besides, this very well describes the situation in the field, which indeed lacks a greater number of reliable studies focusing on (or at least including) the elderly.This highlights the fact that hearing at audiometric frequencies from 9 to 20 kHz deserves further investigation, especially in older adults because the values measured across the studies differed considerably. We could also mention that several other studies reporting that they measured hearing thresholds at high frequencies in the elderly have been published and it can, therefore, appear that our list of studies is not complete.However, it should be noted that those studies used the term "high frequencies" to describe a frequency range of 4-8 kHz, which did not meet our criteria and led to their exclusion from this review. CONCLUSION Significant age-related hearing loss at frequencies above 8 kHz may be observed in people older than 60 years.Currently, not many studies are available that assess extended HFA in individuals older than 60 years; those considering individuals over 70 years of age are even rarer.Extended HFA and hearing thresholds at EHF deserve further investigation as the current estimated reference values for those frequencies in the elderly are burdened with a great amount of uncertainty. Figure 2 . Figure 2. Hearing thresholds audiogram for men and women in age categories of 60-70 years and >70 years, ( 1 A=Hearing thresholds at EHF in men aged 60-70 years across the studies; 2 B=Hearing thresholds at EHF in men aged >70 years across the studies; 3 C=Hearing thresholds at EHF in women aged 60-70 years across the studies; 4 D=Hearing thresholds at EHF in women aged >70 years across the studies). Table 1 . Overview of the Selected Studies Studies Involving Participants Older Than 60 Years Only 17 2004, Stenklev et al17published the results of their study investigating the age-related changes in hearing thresholds in 232 randomly selected elderly (over 60 years of age) individuals, of which 60 were otologically normal.The participants completed CA (frequency range of 0.125-8 kHz) and extended HFA (frequency range of 9-16 kHz).The results were compared to the reference values defined by ISO 7029, Table 2 . EHF Hearing Thresholds of the Elderly in the Included Studies; Median (M) or Mean Values (m) Across Studies BHE, better hearing ear; M, median; m, mean values; NM, not measured; NR, no response.	2024-08-21T06:17:13.551Z	2024-07-01T00:00:00.000	{ "year": 2024, "sha1": "bf7ee40b54ff1c0bfaa831e8f40c09122c936784", "oa_license": "CCBYNC", "oa_url": "https://doi.org/10.5152/iao.2024.231217", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "9caba33490bb7bd7437e5125275fe8e7a00b82aa", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
215773557	pes2o/s2orc	v3-fos-license	Heterologous Expression of a Soybean Gene RR34 Conferred Improved Drought Resistance of Transgenic Arabidopsis. Two-component systems (TCSs) have been identified as participants in mediating plant response to water deficit. Nevertheless, insights of their contribution to plant drought responses and associated regulatory mechanisms remain limited. Herein, a soybean response regulator (RR) gene RR34, which is the potential drought-responsive downstream member of a TCS, was ectopically expressed in the model plant Arabidopsis for the analysis of its biological roles in drought stress response. Results from the survival test revealed outstanding recovery ratios of 52%–53% in the examined transgenic lines compared with 28% of the wild-type plants. Additionally, remarkedly lower water loss rates in detached leaves as well as enhanced antioxidant enzyme activities of catalase and superoxide dismutase were observed in the transgenic group. Further transcriptional analysis of a subset of drought-responsive genes demonstrated higher expression in GmRR34-transgenic plants upon exposure to drought, including abscisic acid (ABA)-related genes NCED3, OST1, ABI5, and RAB18. These ectopic expression lines also displayed hypersensitivity to ABA treatment at germination and post-germination stages. Collectively, these findings indicated the ABA-associated mode of action of GmRR34 in conferring better plant performance under the adverse drought conditions. Introduction Being immobile, plant growth and development are vulnerable to environmental changes [1]. Under overpopulation and climate change pressures, water crisis has become the major concern of many countries around the world and affects multiple aspects of human life including health care, environment and food availability [2][3][4]. In agriculture, various measures have been engaged to minimized yield loss due to limited water supply, including modification of irrigation methods and homozygous transgenic plants identified two independent lines (1.29 and 4.31). These transgenic plants were then subjected to the examination of GmRR34 expression and phenotypic characteristics. Relative expression of GmRR34 under normal growth condition, measured by RT-qPCR, confirmed the transgene expression in the transgenic lines although a non-specific product was also detected in the WT plants at very low expression level ( Figure 1A). It is observed that the amount of GmRR34 transcripts in line 1.29 was nearly double of that in line 4. 31. Under normal conditions, both ectopic expression lines exhibited a reduced growth phenotype when compared with the wild-type (WT) plants. According to the analysis, the transgenic group had smaller average rosette radius and area by approximately 16% and 23%, respectively, in line 1.29, and by around 10% and 16%, respectively, in line 4.31 than did the WT ( Figure 1B,D). Consistently, shorter root lengths by ca. 17% in 1.29 and 13% in 4.31 were also observed in comparison with the WT (Figure 1C,D). Plants 2020, 9, x FOR PEER REVIEW 3 of 18 transformed homozygous transgenic plants identified two independent lines (1.29 and 4.31). These transgenic plants were then subjected to the examination of GmRR34 expression and phenotypic characteristics. Relative expression of GmRR34 under normal growth condition, measured by RT-qPCR, confirmed the transgene expression in the transgenic lines although a non-specific product was also detected in the WT plants at very low expression level ( Figure 1A). It is observed that the amount of GmRR34 transcripts in line 1.29 was nearly double of that in line 4. 31. Under normal conditions, both ectopic expression lines exhibited a reduced growth phenotype when compared with the wild-type (WT) plants. According to the analysis, the transgenic group had smaller average rosette radius and area by approximately 16% and 23%, respectively, in line 1.29, and by around 10% and 16%, respectively, in line 4.31 than did the WT ( Figure 1B,D). Consistently, shorter root lengths by ca. 17% in 1.29 and 13% in 4.31 were also observed in comparison with the WT (Figure 1C,D). Ectopic Expression of GmRR34 Improved Post-Drought Survival Rates in Transgenic Arabidopsis To evaluate the drought resistance mediated by the ectopic expression of GmRR34, drought treatment was applied to evaluate survival ratio of each genotype. After 14 days of water withholding (i.e., soil moisture content (SMC) dropped from 65% to 17%, Figure 2A), both transgenic lines demonstrated their remarkable drought resistance when compared with the WT's capacity. After 3 days of re-watering, the survival ratios were 52% and 53% for ectopic overexpression lines 1.29 and 4.31, respectively, which were significantly higher than that of the WT (merely 28% survived) ( Figure 2B,C). Additionally, examination of water loss rates from detached aerial tissues of transgenic Arabidopsis versus WT plants was conducted. As shown in Figure 2D, the results further illustrated the GmRR34-transgenic plants with better capability of retaining water over the course of dehydration treatment. According to our findings, the degrees of fresh weight (FW) reduction observed in both lines 1.29 and 4.31 were significantly lower than in WT at the same examined time points (e.g., water loss rate of 53% in WT compared with 39% in 1.29 and 45% in 4.31 upon 5-h-dehydration treatment). These results showed that line 1.29 displayed the lowest water loss rate ( Figure 2D). Ectopic Expression of GmRR34 Improved Post-drought Survival Rates in Transgenic Arabidopsis To evaluate the drought resistance mediated by the ectopic expression of GmRR34, drought treatment was applied to evaluate survival ratio of each genotype. After 14 days of water withholding (i.e., soil moisture content (SMC) dropped from 65% to 17%, Figure 2A), both transgenic lines demonstrated their remarkable drought resistance when compared with the WT's capacity. After 3 days of re-watering, the survival ratios were 52% and 53% for ectopic overexpression lines 1.29 and 4.31, respectively, which were significantly higher than that of the WT (merely 28% survived) ( Figure 2B,C). Additionally, examination of water loss rates from detached aerial tissues of transgenic Arabidopsis versus WT plants was conducted. As shown in Figure 2D, the results further illustrated the GmRR34-transgenic plants with better capability of retaining water over the course of dehydration treatment. According to our findings, the degrees of fresh weight (FW) reduction observed in both lines 1.29 and 4.31 were significantly lower than in WT at the same examined time points (e.g., water loss rate of 53% in WT compared with 39% in 1.29 and 45% in 4.31 upon 5-hdehydration treatment). These results showed that line 1.29 displayed the lowest water loss rate ( Figure 2D). Error bars indicate standard errors. Statistically identified differences among three genotypes under the same treatment were indicated by different letters (p-value < 0.05). WW, well-watered and D, 14-day-drought-treated conditions. GmRR34-Ectopic Expression Lines Displayed More Active Enzymatic ROS-Scavenging Activities Endogenous level of hydrogen peroxide (H 2 O 2 ) in WT and transgenic Arabidopsis was investigated to evaluate the degree of cellular damage caused by drought effects. According to our study, the WT plants exposed to drought displayed 3.5-and 2.5-fold higher levels of endogenous H 2 O 2 than the corresponding levels measured in transgenic lines 1.29 and 4.31, respectively ( Figure 3A,B). As antioxidative enzymes are known to be involved in the reactive oxygen species (ROS) detoxification process [15], we further examined the common ROS-scavenging enzymes catalases (CAT) and superoxide dismutase (SOD). The biochemical analyses demonstrated that the transgenic plants were better at detoxifying ROS than WT, since both lines 1.29 and 4.31 showed enhanced CAT and SOD activities, not only under drought but also well-watered conditions, in comparison with the corresponding enzyme activities in the WT ( Figure 3C). Another notice was the more pronounced degree of SOD induction upon the stress application (increased by 31% in line 1.29 and 35% in line 4.31 versus 26% in WT). Additionally, our RT-qPCR results revealed the enhanced expression of genes coding for these enzymes. Relative expression levels of CAT-and SOD-encoding genes, CAT1 (catalase1) and CSD1 (copper/zinc superoxide dismutase 1) [37,38], were found to be substantially higher in the transgenic Arabidopsis when compared with the corresponding levels in the WT plants under the drought treatment ( Figure 3D). Specifically, CAT1 expression was up-regulated by drought by 7.9-fold in line 1. Stress-related Genes with Enhanced Expression in GmRR34-Transgenic Arabidopsis To obtain further molecular insights regarding GmRR34 function in mediating plant resistance to drought, we also examined the expression of three well-known drought-responsive marker genes by RT-qPCR, including RD29A (responsive to desiccation 29A), LEA14 (late embryogenesis abundant 14) and HSP70B (heat shock protein 70B). Statistical analyses of the expression of these genes showed a significantly higher accumulation of their transcripts when compared with those in the WT Stress-Related Genes with Enhanced Expression in GmRR34-Transgenic Arabidopsis To obtain further molecular insights regarding GmRR34 function in mediating plant resistance to drought, we also examined the expression of three well-known drought-responsive marker genes by RT-qPCR, including RD29A (responsive to desiccation 29A), LEA14 (late embryogenesis abundant 14) and HSP70B (heat shock protein 70B). Statistical analyses of the expression of these genes showed a significantly higher accumulation of their transcripts when compared with those in the WT counterparts during water deficit, albeit no differences in the relative expression could be observed under normal conditions. Among these genes, RD29A exhibited the highest up-regulation level upon drought treatment (i.e., 51-and 68-fold induction in 1.29 and 4.31 lines, respectively) while the other two genes LEA14 and HSP70B showed similar increase in expression profile, which was around 20-fold in both transgenic lines ( Figure 4). Plants 2020, 9, x FOR PEER REVIEW 7 of 18 counterparts during water deficit, albeit no differences in the relative expression could be observed under normal conditions. Among these genes, RD29A exhibited the highest up-regulation level upon drought treatment (i.e., 51-and 68-fold induction in 1.29 and 4.31 lines, respectively) while the other two genes LEA14 and HSP70B showed similar increase in expression profile, which was around 20fold in both transgenic lines ( Figure 4). . Three biological replicates were used for each genotype under the same treatment, and statistically identified differences among them were indicated by different letters (p-value < 0.05). Error bars indicate standard errors. GmRR34 Ectopic Expression Lines Displayed Hypersensitivity to ABA Sensitivities of the transgenic plants to abscisic acid (ABA) were examined by various parameters, including evaluation of germination rate, cotyledon development and root elongation of young seedlings upon exogenous ABA application. According to our results, all examined genotypes grown on MS medium without ABA supplementation shared similar rates of germination (approximately 93%). Nevertheless, the two transgenic lines exhibited highly repressed germination rates in comparison with that of WT upon the application of the same ABA concentrations ( Figure 5A). With 0.3 µM of ABA, germination rates of the ectopic expression lines were around 20%-27% lower than that of WT. Under higher exogenous ABA condition (0.5 µM), the transgenic group showed substantial suppression to germination, whereby their germination rates were approximately a half of that of the WT. The recorded proportions of cotyledon greening also consistently illustrated hypersensitivity of GmRR34-transgenic plants to ABA ( Figure 5B). Although comparable percentages of seedlings with greening cotyledon at roughly 93% were observed among the genotypes under the control conditions, there was a decline in number of seedlings with successful cotyledon development upon ABA treatments, and with higher proportions in the transgenic lines in comparison with that of the WT. As shown in Figure 5B, the percentages of plants with cotyledon greening were decreased by 2.5-and 3.5-fold in line 1.29, and 1.9-and 3.7-fold in line 4.31 in the media supplemented with 0.3 and 0.5 µM ABA, respectively. In agreement with these data, inhibition of root elongation in the presence of ABA was more severe in the transgenic group (i.e., reduced by 16%-32% to 72%-73% in the transgenic lines versus 11% to 39% in WT at 0.3 and 0.5 µM ABA, respectively) ( Figure 5C and Supplementary Figure S1). Additionally, the expression of several key genes belonging to ABA biosynthesis and signaling pathway, including NCED3 (nine-cisepoxycarotenoid dioxygenase 3), OST1/SnRK2.6 (open stomata 1/snf1-related protein kinase 2.6), ABI5 (ABA insensitive 5) and RAB18 (responsive to ABA 18), was evaluated by RT-qPCR. The data showed that expression of these genes was enhanced in the transgenic lines. Under the drought conditions, apart from NCED3 for which solely the 1.29 ectopic expression line exhibited significantly higher expression level than did the WT, the expression levels of other examined ABA-responsive genes in both overexpression lines were remarkably higher by an average of 40% in comparison with corresponding levels in WT plants experiencing the same drought treatment ( Figure 5D). . Three biological replicates were used for each genotype under the same treatment, and statistically identified differences among them were indicated by different letters (p-value < 0.05). Error bars indicate standard errors. GmRR34 Ectopic Expression Lines Displayed Hypersensitivity to ABA Sensitivities of the transgenic plants to abscisic acid (ABA) were examined by various parameters, including evaluation of germination rate, cotyledon development and root elongation of young seedlings upon exogenous ABA application. According to our results, all examined genotypes grown on MS medium without ABA supplementation shared similar rates of germination (approximately 93%). Nevertheless, the two transgenic lines exhibited highly repressed germination rates in comparison with that of WT upon the application of the same ABA concentrations ( Figure 5A). With 0.3 µM of ABA, germination rates of the ectopic expression lines were around 20%-27% lower than that of WT. Under higher exogenous ABA condition (0.5 µM), the transgenic group showed substantial suppression to germination, whereby their germination rates were approximately a half of that of the WT. The recorded proportions of cotyledon greening also consistently illustrated hypersensitivity of GmRR34-transgenic plants to ABA ( Figure 5B). Although comparable percentages of seedlings with greening cotyledon at roughly 93% were observed among the genotypes under the control conditions, there was a decline in number of seedlings with successful cotyledon development upon ABA treatments, and with higher proportions in the transgenic lines in comparison with that of the WT. As shown in Figure 5B, the percentages of plants with cotyledon greening were decreased by 2.5-and 3.5-fold in line 1.29, and 1.9-and 3.7-fold in line 4.31 in the media supplemented with 0.3 and 0.5 µM ABA, respectively. In agreement with these data, inhibition of root elongation in the presence of ABA was more severe in the transgenic group (i.e., reduced by 16%-32% to 72%-73% in the transgenic lines versus 11% to 39% in WT at 0.3 and 0.5 µM ABA, respectively) ( Figure 5C and Supplementary Figure S1). Additionally, the expression of several key genes belonging to ABA biosynthesis and signaling pathway, including NCED3 (nine-cis-epoxycarotenoid dioxygenase 3), OST1/SnRK2.6 (open stomata 1/snf1-related protein kinase 2.6), ABI5 (ABA insensitive 5) and RAB18 (responsive to ABA 18), was evaluated by RT-qPCR. The data showed that expression of these genes was enhanced in the transgenic lines. Under the drought conditions, apart from NCED3 for which solely the 1.29 ectopic expression line exhibited significantly higher expression level than did the WT, the expression levels of other examined ABA-responsive genes in both overexpression lines were remarkably higher by an average of 40% in comparison with corresponding levels in WT plants experiencing the same drought treatment ( Figure 5D). Discussion Previous studies have provided strong lines of evidence for the participation of the soybean type-C RR gene GmRR34 in drought stress responses [35,36]. According to these reports, expression of GmRR34 was markedly up-regulated in root and/or shoot tissues under drought [36] and dehydration [35]. Furthermore, sequence analysis of GmRR34 recognized a cis-motif MYCR (myelocytomatosis-related protein recognition site) in its 1000-bp promoter region [30]. MYC transcription factor (TF) family has been known to act in drought-responsive pathway in an ABA-dependent manner [17,39]. These findings suggested that GmRR34 might contribute a major role in soybean response to water deficit, thus conferring the plant resistance to drought stress. In this study, we aimed to explore the biological functions of GmRR34 in relation to drought responses by using the heterologous system. The expression of GmRR34 in the two homozygous independent transgenic Arabidopsis lines that we successfully screened out was confirmed by RT-qPCR ( Figure 1A). The findings showed that the expression levels of GmRR34 in lines 1.29 and 4.31 were 809-and 452-fold higher, respectively, than the expression level detected in WT plants, when using the GmRR34 primers. Previous bioinformatic analyses revealed that a GmRR34 homolog, AtRR24, was found in Arabidopsis genome [30]. This helps explain why a non-specific product has been amplified in the WT plants. In addition, our findings indicated that the ectopic expression of GmRR34 resulted in retarded growth in the transgenic plants under normal growth conditions ( Figure 1B-D). Previous studies also reported on similar phenotypic alteration in relation to improved drought tolerance/resistance when using constitutive 35S promoter to drive the expression of transgenes in various transgenic species, including Arabidopsis [40][41][42], tobacco [43,44], and tomato [45,46]. Results from survival test illustrated that the transgenic plants ectopically expressing GmRR34 acquired better performance against 14-day-drought treatment than their WT counterpart, as the number of transgenic plants in each genotype that could continue their life after drought treatment and re-watering was nearly double that of the WT ( Figure 2B,C). The transgenic lines displayed lower water loss rates from detached aerial parts than their WT counterpart, which could be associated with their smaller rosette phenotype ( Figure 1B-D and Figure 2D). In plant adaptation to drought, inhibition in shoot and leaf growth and thus reduced transpiration was promoted due to alterations in concentrations of various phytohormones, including ABA, CK and gibberellin [47][48][49][50][51]. Therefore, better cellular water conservation is an important attribute to plant survival under water-deficit conditions [41,52]. Examining water loss rate from detached leaves/rosettes has been widely used as important, reliable indicator to rapidly predict the plant water status and performance under drought conditions [53][54][55]. A great number previous studies included this tool in evaluating the performance of various transgenic plants, with the tracking of water loss rates over the course duration of 3 h [56,57], 4 h [58,59], 5 h and longer [52,60]. Upon being excised, water loss is indicated to be achieved via stomatal transpiration within the first hour followed by cuticular transpiration and residual transpiration (i.e., from incompletely closed stomata) in subsequent hours [61]. From our data, in the first hour, WT transpired more than transgenic plants, and line 1.29, with higher GmRR34 expression, transpired the lowest. This could indicate a direct or indirect relationship with stomatal conductance regulation. Additionally, differences after the first hour tend to remain, indicating a decreased cuticular transpiration in the transgenic plants ( Figure 2D). It is also worth noting that depending on duration and intensity of a certain drought condition, different responsive mechanisms might be used by the plants. Drought stress often induces oxidative stress due to a burst production of ROS, which are the causes for cellular structural damage and biological function impediment [62,63]. Despite being a toxic by-product of cellular metabolism activities, under normal growth conditions, ROS levels are delicately maintained at a steady and low concentrations by ROS-scavenging activities as they contribute major parts in various signaling pathways [15,64,65]. During drought events, in addition to the electron transport chain activities in the chloroplast, photorespiration which is caused by stomatal closure is another source for extensive ROS production, especially H 2 O 2 [66,67]. Therefore, endogenous H 2 O 2 levels and detoxification activities by enzymes are commonly measured as important parameters in drought tolerance studies [40,42]. As 14-day-drought duration used in the survival test would be too severe and thus cause difficulties in analyzing biochemical data due to protein degradation and plant death, we determined to use a shorter stress duration (i.e., 10 days) for these assays. In agreement with higher survival ratios and lower water loss rates, both GmRR34-transgenic lines appeared to be "less stressed" with lower endogenous H 2 O 2 contents and higher activities of CAT and SOD enzymes ( Figure 3B,C). Furthermore, expression analyses of genes encoding CAT and SOD revealed significantly higher transcript abundance in the ectopic expression lines ( Figure 3D). These data indicate that the lower H 2 O 2 contents in lines 1.29 and 4.31 versus the WT could be the result of higher CAT activities and CAT1 expression in the former group [68]. CAT1 and CSD1 are known to encode ROS-scavenging enzymes CAT and SOD, respectively, which catalyze the conversion of H 2 O 2 and superoxide (i.e., another type of ROS), accordingly [37,38]. These results altogether imply that the transgenic plants were better protected from damages by ROS accumulation. As there might be other gene members of CAT and SOD families which are also involved in mediating plant response to drought [37,69], it is not surprised when the biochemical and expression data were not entirely correlated. In Arabidopsis, three CAT members (CAT1, CAT2, and CAT3) have been reported to participate in decomposition of H 2 O 2 and regulating ROS homeostasis [70]. While CAT1 expression has been linked with various stress conditions such as drought, in normal growth, the majority of CAT activities in controlling ROS homeostasis are the results of CAT2 and CAT3 [37,70]. Similarly, detoxification of superoxide was found to be the result of seven SOD enzymes, three of which were copper/zinc superoxide dismutases (CSD1, CSD2, and CSD3) [69]. Therefore, investigation on expression of other CAT-and SOD-encoding gene members under drought context should be conducted in future studies. Three well-known stress-responsive genes, RD29A, LEA14, and HSP70B, were also selected to examine their expression in our study to gain insights to the molecular function basis of GmRR34. Expression patterns of these genes have been considered important indicators for drought-responsive studies [71,72]. As shown in Figure 4, under drought condition, expression levels of all these genes were significantly higher in the transgenic lines compared with those in WT by 30% (HSP70B) to nearly 50% (RD29A and LEA14). Studies have reported rapid induction of the stress marker gene RD29A under various abiotic stresses, such as water deficit, and particularly in stress-tolerant plants, thus emphasized its role as indicator for resistance toward drought [71,73,74]. Herein, in correlation with higher resistance to drought of transgenic plants (Figure 2), RD29A expression was also found to be remarkedly induced, suggesting RD29A is one of GmRR34 target for mediating drought responses, through a direct and/or indirect manner. LEA14 and RAB18 belong to LEA family, a group of proteins that play important roles in protecting cellular proteins and membrane stabilization under osmotic stress conditions [72,75]. On the other hand, HSP70B, of which expression was found to be highly expressed under extreme temperature conditions [76], belongs to a major HSP family and acts as a molecular chaperone and folding catalyst [77]. HSP70B has been previously demonstrated to be regulated by drought-related TFs, including MYB21 [78]. The fact that HSP70B was recorded to be noticeably induced upon drought in our study suggests its conserved roles in maintaining protein structures not only under heat stress but also under water stress conditions. Comprehensively, higher induction of the studied stress-responsive genes in transgenic plants ectopically expressing GmRR34 implied the beneficial role of GmRR34 in plant resistance to drought. According to our findings, both lines of transgenic Arabidopsis were hypersensitive to the presence of exogenous ABA, illustrated by substantially lower rates of germination and seedlings with normal cotyledon development as well as reduced root growth ( Figure 5A-C). Ever since its discovery, ABA has been intensively studied and its functions have been linked to various stress responses, from water relation such as drought, cold or heat to wounding or pathogen infection, by inducing stomatal closure or regulating the expression of ABA-responsive genes [17,[79][80][81]. Thus, such ABA-hypersensitive traits imply ABA-dependent activities of GmRR34 and possibly faster responses to osmotic changes in the transgenic plants. Known ABA-mediated stress regulations include guard cell responses or production of osmo-compatible solutes, which play major roles in retaining water from evaporation and maintaining cellular water potential [72,79,82]. Similar to findings for the expression pattern of RD29A, LEA14, and HSP70B, expression levels of ABA-related genes were comparable under normal condition among all genotypes but at substantially higher levels in the transgenic plants under drought (Figures 4 and 5D). This observation could be explained that the constitutive expression of GmRR34 under non-stressed condition was not sufficient to trigger alteration in expression of these studied genes, whereas upon the stress challenge, activities of these genes could be under the control of a complicated multi-component mechanism related to drought response, suggestively including GmRR34, that are worthy to be elaborated more in the forthcoming studies. The higher expression of ABA-related genes in the ectopic expression lines implies certain advantages acquired by the plants under drought conditions. NCED3, which codes for 9-cis-epoxycarotenoid dioxygenase, is a key enzyme for ABA biosynthesis [83]. Among the seven cloned AtNCEDs in Arabidopsis, NCED3 was the only gene induced by drought [83]. From our results, the fact that expression of NCED3 was highly induced along with the outstanding expression of GmRR34 in transgenic line 1.29 under drought conditions suggests a possible relation between GmRR34 and ABA biosynthesis ( Figures 1A and 5D). Meanwhile, OST1 had been documented by numerous studies and recognized as a core component of ABA signaling. In this pathway, its functions are associated with the regulation of stomatal closure by phosphorylating various membrane proteins and ion channels such as plasma membrane intrinsic proteins (PIPs), slow anion channel-associated 1 (SLAC1) and potassium channel in Arabidopsis thaliana 1 (KAT1) [79,[84][85][86]. The other two downstream components of the ABA-signaling pathway, ABI5 and RAB18, are ABA-induced genes and have been demonstrated to function in various abiotic stresses including water deprivation [87,88]. ABI5 is a basic leucine zipper TF, playing key roles in abiotic stress response through regulating gene expression of various downstream targets [88], whereas, RAB18 is a functional protein (i.e., dehydrin), which is responsible for cellular protection under water deficit conditions [87,89]. Furthermore, a previous study demonstrated that CAT1 could be the downstream gene regulated by ABI5 [90], whereby their expression upon drought treatment were both up-regulated ( Figures 3D and 5D). Taken these results with lower water loss rates (importantly within the first h of dehydration) observed in the transgenic Arabidopsis together, it is suggested that GmRR34 enhances drought resistance of the transgenic plants in an ABA-dependent manner, at least by promoting the closure of stomata under water-deficit condition. Plant Materials WT Arabidopsis (Col-0) were used as control plants and materials for generating transgenic plants. GmRR34 (Glyma03g32720) cDNA [35] was inserted into plasmid pBI121 by replacing the selective marker gene β-glucuronidase (GUS) and under the regulation of constitutive Cauliflower mosaic virus (CaMV) 35S promoter. The recombinant vector was transferred into Agrobacterium tumefaciens strain EHA101 for generating transgenic plants by Agrobacterium-mediated floral dip method [91]. The transgenic plants were then screened for independent homozygous progenies based on the ratio of kanamycin-resistant over kanamycin-sensitive phenotypes (applied antibiotic at concentration of 30 mg L −1 ) over three successive generations from T1 (with 7 independent lines) to T3 according to the Mendelian genetic laws [92]. Growth Conditions The methods for seed sterilization, germination and normal growth conditions were described in Nguyen et al. (2019) [42]. In brief, the seeds of studied lines were sterilized by ethanol and sodium hypochlorite prior to germination on Murashige and Skoog (MS) medium (1% sucrose, 0.8% agar). Treated seeds were stratified in the dark for 2 days before proceeding to normal growth condition using controlled facility (22 • C, 60% humidity and 16/8 light-dark cycle). In all experiments, except ABA sensitivity assay, the seeds were germinated and grown on MS medium for ten days. These seedlings were then transferred to soil (Tribat, Saigon Xanh Biotechnology Ltd. Company, Ho Chi Minh City, Vietnam), and allowed to grow normally in the same growth room for additional fourteen days before being subjected to the assays. Survival test and water-loss assay were performed separately. For the analysis of hydrogen peroxide content, enzyme activities, and expression of drought-related genes, the same set of plants were used. Phenotypic Analyses of Ectopic Expression Lines under Normal Conditions Root length of 24-day-old plants was measured by taking the plants out of soil without damaging the root system [42]. Length from the longest leaf of each plant was used to determine the maximum rosette radius, using the Image-J software (https://imagej.nih.gov/ij/) [41]. Arabidopsis rosette area measurement was carried out following the method described in a previous study [93] using Photoshop CC 2019 (Adobe, San José, CA, USA). Each experiment was performed with ten biological replicates. Survival Test The plants growing in tray system were subjected to drought stress by withholding water until the relative soil moisture content (SMC) dropped below 20%, using moisture meter (TK-100G, Yieryi, Guangdong, China) to measure randomly 6 different soil positions for each treatment per timepoint. Other growing factors were maintained the same as specified in Section 4.2. Following the drought treatment, the treated plants were re-irrigated for 3 days before recording the survival ratio of each genotype, of which the criterion was the capacity of developing new green tissues [42]. The experiment was triplicated, with 20 plants per replicate. Determination of Excised-Leaf Water Loss Rate Aerial part from individual 24-day-old-plants were excised for performing water loss assay. In brief, the tissues were dehydrated at room temperature condition (25 • C) and their FWs were recorded at 1-h-intervals over a period of 5 h. Reduction in FWs during the studied course was used to estimate the water loss rates from the samples [94]. Nine plants of similar phenotype were used in this experiment for each genotype. Evaluation of H 2 O 2 Content and ROS-Scavenging Enzymatic Activities Leaf tissues from plants that had been subjected to 10-day drought treatment were used for analyses (0.2 g leaves per plant as a replicate, three biological replicates). Endogenous H 2 O 2 content was determined following Patterson's procedure [95] with minor modification by using 0.1% titanium(III) sulfate and 20% sulfuric acid in preparing reaction solution. Previously described methods were followed for quantification of soluble proteins [96], evaluation of SOD [97] and CAT [98] enzymatic activities. The enzymes were extracted in potassium phosphate buffer (1 M, pH 7.8) containing EDTA 0.1 M and 2% polyvinylpyrrolidone (M.W. 8000). Bovine serum albumin (Sigma, Saint Louis, MO, USA) was used to establish the standard curve for protein quantification. Expression Analysis of Transgene and Drought Response-Related Genes To analyze expression of drought-related genes by RT-qPCR, the shoot tissues without inflorescence (3 biological replicates) of both 10-day-drought-treated seedlings and of the control plants grown under normal conditions were collected. For checking transgene expression, well-watered samples of transgenic plants were used, along with WT as negative control. Total RNA isolation, cDNA synthesis and RT-qPCR were conducted following the methods previously described in Hoang et al. (2020) [40]. Primers used in this experiment were obtained from previous studies (Supplementary Table S1). Relative expression level was analyzed by 2 −∆Ct method [99], and Actin2 was used as the reference gene [100]. ABA Sensitivity Assays Plants were cultivated on MS medium without ABA and with ABA at concentration of 0.3 or 0.5 µM under normal growth conditions. To figure out the germination rates, the seeds with emerged radicles within the examined population were counted after 2 days of stratification followed by 3 days of incubation. To determine the rates of successful cotyledon development, seedlings with green cotyledons were counted after 7 days of incubation. These procedures were performed using methods described in previous study [101]. Each experiment was triplicated, with 100 seeds per replicate. For evaluation of root elongation, seeds of WT and transgenic lines were germinated on MS medium and allowed to grow for 4 days. After that, the seedlings were transferred to MS plates, supplemented with different concentrations of ABA (0, 0.3, and 0.5 µM) and let grow vertically for 8 days prior to the measurement of root length [102]. Ten biological replicates were used for each genotype for determining of root growth capacity under ABA treatment. Statistical Analysis To identify statistically significant difference among genotypes under the same treatment, data were analyzed using one-way ANOVA (Tukey's honestly significant difference-HSD test). Conclusions The combined results from this study have demonstrated that GmRR34 plays as a positive regulator in mediating plant response to water deficit condition in an ABA-dependent manner, suggestively at least in Arabidopsis and/or soybean plant systems. Its biological functions have been found to be involved in various processes, including water retention ability, antioxidant enzyme activities, and regulating a number of drought-related genes. Conflicts of Interest: The authors declare no conflict of interest.	2020-04-16T09:09:04.539Z	2020-04-01T00:00:00.000	{ "year": 2020, "sha1": "53bad6424d52ddeacc3b9dce688b76dc14c89554", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2223-7747/9/4/494/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "19d366509858d32ca5fcd392fc863a5c8203ed97", "s2fieldsofstudy": [ "Environmental Science" ], "extfieldsofstudy": [ "Medicine", "Biology" ] }
234930514	pes2o/s2orc	v3-fos-license	Pooled Data Analysis of Tourism Industry in Special Region of Yogyakarta Tourism industry is one of sectors which is relied by government as regional acceptance post, therefore, the government of Special Region of Yogyakarta is requested to be able to dig and manage tourism potencies that it has. Tourism industry can be developed as an effort to get source of funds through new innovation in having effort to finance regional expense through retribution obtained from every tourism attraction in each region. The objective of this research is to estimate tourism industry, which is in this case, they are variables of the number of tourism attractions, the number of hotels and GRDP as independent variables towards variable of tourism retribution as dependent variable with the objects of research are all areas of districts/city in Special Region of Yogyakarta. Data used in this research were data of time series which in this case, they were from researches is 2010-2018 and data of cross section which were data of tourism industry for all areas of districts/city in Special Region of Yogyakarta. The analysis method used was analysis of pooled data which combine data of time series and cross section, and in testing, it used software of Eviews 9.0. This research model is able to describe variables which influence tourism retribution as much as 91,63%. It can be proved on testing fixed effect on R2 as much as 0.916325 in meaning that the independent variables, the number of tourism attractions, the number of hotels and GRDP, are able to desribe the dependent variabel, tourism retribution, as much as 91,63%. I. INTRODUCTION Tourism is one of service industry activities that become Indonesia's mainstay in improving foreign exchange in sector of non oil and gas. In essence, natural wealth, art and culture, and diversity of tourism potencies such as various facilities owned by the region can become basic capital of development and sustainable tourism development. The development of tourism has important role as development and economic growth in creating healthy and dinamic condition through managing business activities and tourism in region [1,2]. Tourism industry is one of fields that has a big opportunity as a factor to promote national economy today [3]. This sector is expected to be able to play a role as the mainstay source of foreign exchange income, and a field that is able to create a lot of jobs as well as improve investment. Efforts conducted by government in increasing tourism field are makes plans and various policies, one of its policies is by digging, making inventory and advancing every tourism attraction existed as a magnetism for tourists. Tourism field has many potencies that can be utilized, one of them is as an effort to improve Own Source Revenue by taking retribution of tourism attraction [4,5]. Special Region of Yogyakarta is an region that has various tourism places, either natural tourism, artificial tourism or historical tourism. They have thier own magnitism compared to other regions. Tourism industry in Special Region of Yogyakarta is expected to be able to give sufficient income contribution for regional government because the number of tourists visiting to Yogyakarta are getting up. Therefore, researcher proposes the research title about tourism industry in Special Region of Yogyakarta using approach of pooled data. The problems in this research can be formulated as follows (i) is there any influence of the number of tourism attractions towards tourism retribution in Special Region of Yogyakarta?; (ii) is there any influence of the number of hotels towards tourism retribution in Special Region of Yogyakarta?; (iii) Is there any influence of GRDP towards tourism retribution in Special Region of Yogyakarta?; (iv) Is there any influnce simultaneously among the number of tourism attractions, the number of hotels and GRDP towards tourism retribution in Special Region of Yogyakarta?. While the objectives of the research are (i) estimating the influence of the number of tourism attractions towards tourism retribution in Special Region of Yogyakarta; (ii) estimating the influence of the number of hotels towards tourism retribution in Special region of Yogyakarta; (iii) estimating the influence of GRDP towards tourism retribution in Special Region of Yogyakarta; (iv) estimating simultaneously the number of tourism attractions, the number of hotels, GRDP towards tourism retribution in Special Region of Yogyakarta. The researches about tourism industries with analysis of panel data have not much been carried out, therefore in this research, it is more emphasized on the approach of panel data for tourism industry in Special Region of Yogyakarta. A. Tourism History Tourism has been known in the world since prehistoric times, however the definition of tourism at that time is not similar today (modern time). Long time ago, the world nations such as Sumeria, Phoenisia, until Roman, already travelled, but the aim was for trading, improving knowledge of living science, or political science. Then, after modernization is widespread all over the world, especially after revolution of industry in England, then travellers are come up alternately to do travelling as we know today [6]. Whereas in Indonesia itself, tourism has been known since kingdom era which dominated archipelago area, although at that time they were concerned for dominating each other, however it could not be undeniable that there was cultural exchange among areas. Indonesia modern tourism has been known since Dutch colonialism era in Indonesia. Through Vereeneging Toesristen Verker (VTV), it was an institution or official tourist bureau. The position of VTV besides as a tourism institution also acted as tour operator or travel agent. At that time, tourism institution was founded by Dutch just prioritizing for only white tourists, while for natives themselves were given limitation as it was done for other sectors. After independence, Indonesia tourism, little by little, showed enhancement. From the period of Repelita I to Repelita IV tourist in Indonesia were getting up drastically, even exceeded the target, that were 11.626.000 tourists which was in the beginning that were targeted only 3.000.000 people. The term of tourism was firstly introduced by two cultural practitioners in 1960's, by Moh. Yamin and Prijono. Both of them gave idea to government to change the term of tour in order to be appropriate to special language of archipelago [7]. B. Tourism Defines tourism as an economical activity which has to be seen from two sides namely demand side and supply side [8,9]. Furthermore, it is stated that the successfulness in developing tourism in an area depends on ability of designer in integrating both sides equally into a tourism development plan. From the demand side, for instance, it has to be identified potencial market segments for the area concerned and the main attraction factors for area of tourism destination. This description shows stategical value of tourism plan, therefore tourism can give prior benefit for public welfare [10]. According to [11], tourism is a tour activity done for a while from original residence to destination area with the reason not to stay or work but only for having fun, fulfilling his/her curiosity, spending his/her leisure time or holiday as well as other objectives. However, basically, tourism is an indication of people movement temporaly and spontaneously in order to fulfill needs and curiosity [12]. Tourism attraction is a place or natural condition having tourism source that is built or developed so that it has attractions and makes tourists visit [13,14]. According to [15], an area becomes good tourism destination, it has to develop three things so that it attracts to be visited, namely:  Something to see, it means that there is something to see, in this case the tourism attraction differs from other places (having own uniquiness). Besides that, it also needs to concern about tourism attractions that become as entertainment when people visit.  Something to buy, there is something interesting, something special to buy to become souvenir to bring home so that in that area, there are facilities to do shopping that provide souvenir and other handicrafts as well and it has to be supported by other facilities such as money changer and bank.  Something to do, there are activities that can be done in that place which can make vistitor feel at home. From the description above, it can be concluded that a good and interesting tourism place to be visited must have natural beauty and also have uniquiness and attraction to be visited and also supported other facilities to be enjoyed at the time the visitors visit [16]. C. Tourist Tourist is a person motivated by something or various needs to do travel and stop by outside his/her residence for more than 24 hours with no aim to do job [17]. Person who is called tourist, based on constitution No. 9, year of 1990 about tourism is tour done voluntary, tour to other place outside his/her area/country of residence, temporary and staying at least one night, not for work, the main purpose is to have a leisure, holiday, health, religious and do sport, business visit, family visit, assignment and attend conference [18,19]. D. Gross Regional Domestic Product (GRDP) The important indicator to be able to know economical condition of a region in certain period is by using data of Gross Regional Domestic Product (GRDP), it can use on the basis of price applied or constant price. According to [20], economic growth is the increasing of per capita output in long term, the emphasize is on three aspects, namely process, percapita output, as well as long term. Economic growth is a process, not only shortly economic picture. Regional development as well as sectoral development must be implemented in line so that sectoral development in regions is able to functionate in accordance with their potencies as well as region priorities. Gross Regional Domestic Product (GRDP) is the number of additional value that is produced for all business areas and services in an area, applies the number of all good values and end services produced by all economic unit, GRDP itself can be interpreted as the number of additional values that are produced by all business units or the number of all good values and services by all economic unit in an area [21]. Economic growth is the development of public welfare which is measured with the magnitude of pre-capita Gross Regional Domestic Product (per-capita GRDP) growth [22]. The higher the value of GRDP in an area, thus it shows the higher level of economic growth as well as it reflects that the region experiences progress in its economy. Basically, economic growth in a region can occur when endogenous determinants (factors inside the region) or exogenous (factors outside the region) are involved and combined. The approach Advances in Economics, Business and Management Research, volume 169 usually used in describing regional growth is by using models of macro economy [23]. A. Type and Data Resource Data used were secondary data with time span of 14 years, from 2010 to 2018 for districts/city in Special Region of Yogyakarta. The secondary data that were collected involving the data of the number of tourism attractions, the number of hotels, GRDP and total of tourism retributions, those data were published by publication of Government Tourism Office and Central Bureau Statistics of Special Region of Yogyakarta. B. Regression Model of Pooled Data This research used quantitative approach with method of pooled data, namely combining data of time series and cross section. This analysis described relation among dependent variable, tourism retribution and independent variables,the number of tourism attractions, the number of hotels and GRDP. The model made in this research used equation of function as follows: Tourism retribution = F (the number of tourism attractions, the number of hotels, GRDP) (1) Then the equation above put in the form of equation model of pooled data with logarithm as follows : This method is the main method that is used with estimating pool data by combining data of cross section and time series without seeing the difference between time and individu that will obtain a lot of observation. Basic assumption on the approach of PLS are that values of intercept and slope between unit of cross section and time series are the same.  Fixed Effect Model (FEM) Fixed effect model (FEM) is simple method to know the difference between object and constant among objects. It is assumed that intercept among individu are different but the slope are still the same among individu.  Random Effect Model (REM) According to [24] describes that dummy variable is put in into model of random effect in order to represent our nescience about the real model. But this brings the consequences of reduced freedom degree which in turn reducing parameter efficiency. This problem can be solved by using error term which is known as method of random effect. C. Model Selection To test compatibility and goodness of model from three models on model estimation technique with pooled data using Chow test and Hausman test.  Chow test Chow test is called as testing F-statistics that test is to choose model used to compare model of common effect with fixed effect. When value of chow statistic (F-stat) of test result is bigger than F-table, thus it is enough to do rejection towards zero hypothesis so that model used is model of fixed effect.  Hausman test Hausman test is used to select the most appropriate model, whether the most appropriate model uses fixed effect or random effect. Determination result of value of Hausman test is gained by comparing value of chi-square statistic with critical value of chi-square, degree of freedom (df) used is as many as k (number of independent variables). The hypothesis is as follows: If statistics value of Hausman test is bigger than its critical value, thus it rejects Ho and the model used is model of fixed effect, if statistics value of Hausman test is smaller than the ctitical value, thus it accepts Ho and the model used is random effect. A. Model Estimation Testing estimation model is done to find out the suitable model used in analysis of econometrics model. Model selection in this research is conducted by using regression analysis of pooled data by using three models, namely Source: secondary data processed, 2020 Result of Chow Test can be known that cross section F is as much as 11.892102 with probability of 0.0000 and it is significant on α = 5%. Because probability of cross section F is significant on α = 5%, therefore decision making of model used is Fixed Effect.  Hausman Test The result of significance test which is comparation between models of Fixed Effect and Random Effect can be seen as follows: Source: secondary data processed, 2020 From that test result can be known that cross section random is as much as 12.298845 with probability as much as 0.0000 and it is significant on α = 5%. Because probability of cross section random is significant on α = 5%, therefore the decision making of the model used is Fixed Effect. B. Classical Assumption Test Classical assumption test in this research involves normality test, multicollonearity test and heteroscedasticity test.  Normality Test Normality test is to test whether in the regression model, independent variables and dependent variabel are distributed normally or not. Decising making with Jarque-Bera Test or J-B Test which is value of > 5% , thus those variables are distributed normally. Based on normality test on probability as much as 0.759222 shows that probability is > 5%, therefore it can be concluded that the data are distributed normally.  Multicollinearity Test Multicollinearity test is to test whether the regression model occurs correlation among independent variables or not. According to [25] if coefficient of corelation among independent variables is more than 0.8, so it can be concluded that the model has multicollinearity problem. Vice versa, if correlation coefficient among variables is less than 0.8, thus the model is free from multicollinearity problem. Based on the test result of multicollinearity, all correlation coefficients are less than 0.8, therefore it can be concluded that the model is free from problem of multicollinearity.  Heteroscedasticity Test Heteroscedasticity test is to test whether in the regression model, it occurs varian inequality from residual of one observation to other observations. Based on result of heteroscedasticity test with Breusch Pagan LM above, it shows the probability more than 0.05. It means that there is no heteroscedasticity in this model. C. Statistical Analysis Based on the selection of model done, thus model of fixed effect is selected to become the best model to estimate.  Test of (R 2 ) Coefficient of determination (R 2 ) is as much as 0.916325, it means that the independent variables, namely the number of tourism attractions, the number of hotels and GRDP are able to describe the dependent variable, that is tourism retribution as much as 97.70%. While the rest 2.30% is described by other variables outside the model.  F test F test is used to know whether variables together give effect to the dependent variable or not. Based on the test result, it is obtained F-statistic as much as 36.82982 with probability F-statistic as much as 0.000000 < α 5%, which means Ho is rejected. The independent variables, which are the number of tourism attractions, the number of hotels and GRDP together have significant effect towards the dependent variable, that is tourism retribution.  T test T test is used to know whether or not there is the effect of the dependent variable towards the independent variables individually and it is considered that the dependent variable is more constant. Based on test result above, thus the decision making is as follows: a. Testing towards LX1 variable Based on result of data calculation uses program of Eviews 9, thus it is obtained the result that significant value is as much as 0.0011, which means that LX1 variable has significant effect towards Y because significant value is smaller than 0.05 with positive direction of relation. b. Testing towards LX2 variable Based on result of data calculation uses program of Eviews 9, thus it is obtained the result that significant value is as much as 0.0112, which means that LX2 has significant effect towards Y because significant value is smaller than 0.05 with positive direction of relation. c. Testing towards LX3 variable Based on result of data calculation uses program of Eviews 9, thus it is obtained the result that significant value is as much as 0.0005, which means that LX3 variable has significant effect towards Y because significant value is smaller than 0.05 with positive direction of relation. Based on regression equation, hence it can be described that regression coefficient value of the number of tourism attractions as much as 0.912079 shows that when the number of tourism attractions increase as many as 1 unit so tourism restribution is getting up as much as 0.912079 rupiah if other independent variables are in constant condition. The positive coefficient regression means the higher the percentage value of the number of tourism attractions, thus tourism retribution is increasing. Regression coefficient value of population density as much as 1.690465 shows that when the number of hotels increase as many as 1 unit, hence tourism restribution is increasing as much as 1.690465 rupiah if other independent variables are in stable condition The positive regression coefficient means the higher the number of hotels, thus tourism restribution is going up. Based on the regression equation, it can be described that regression coefficient value of GRDP as much as 0.626479 shows that when GRDP is up as many as 1 unit, hence tourism restribution is increasing as much as 0.626479 rupiah if other independent variables are in stable condition. The positive coefficient regression means the higher GRDP the higher the tourism retribution. E. Discussion The effect of the number of tourism attractions towards retribution income in 5 districts/city in Special Region of Yogyakarta. Based on analysis result can be explained that variable of the number of tourism attractions has positive and significat influence with coefficient value as much as 0.912079 towards tourism restribution of districts/city in Special Region of Yogyakarta from 2010 to 2018. It shows that if tourism attraction in a region has increasing as many as 1 unit of tourism attractions, thus it will increase retribution income in Special Region of Yogyakarta as much as 0.912079 rupiah. This result is in accordance to research hypothesis statement that there is positive effect towards tourism retribution income which is seen from the number of tourism attractions of districts/city in Special Region of Yogyakarta during 2010 until 2018. The same statement is also said by [26] finding that there is an effect of the number of tourism attractions towards tourism restribution income. From the regression calculation result, as shown on equation above, shows consistency towards theory that the number of tourism attractions give positive sign. This is in line with the theory mentioning that the existance of the number of tourism attractions can increase regional income of tourism sector in the Special Region of Yogyakarta, either through regional tax or regional retribution. The effect of the number of hotels towards retribution income in 5 districts/city in Special Region of Yogyakarta. Based on analysis result, it can be described that variable of the number of hotels with coefficient value as much as 1.690465 is significant towards tourism restribution income of districts/city in Special Region of Yogyakarta from 2010 to 2018. It shows that if the number of hotels is increasing as many as 1 unit, hence it will develop retribution income of districts/city in Special Region of Yogyakarta as much as 1.690465 rupiah. This result is in accordance to the research hypothesis stating that there is positive effect of the number of 5 districts/city restribution in Special Region of Yogyakarta. The effect of GRDP towards retribution income in 5 districts/city in the Special Region of Yogyakarta. Based on analysis result, it can be described that variable of GRDP with coefficient value as much as 0.626479 is significant towards tourism restribution income of districts/city in Special Region of Yogyakarta from 2010 to 2018. It shows that if GRDP is getting up as many as 1, thus it will improve retibution income of districs/city in Special Region of Yogyakarta as much as 0.626479 rupiah. This result is suitable with the research hypothesis stating that there is positive effect of the number of 5 districts/city retribution in Special Region of Yogyakarta.	2021-05-22T00:04:59.177Z	2020-01-01T00:00:00.000	{ "year": 2020, "sha1": "40ee379fb33c1cc21149c6074ac3d9a1cda7ec23", "oa_license": "CCBYNC", "oa_url": "https://www.atlantis-press.com/article/125953901.pdf", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "381dfb2cd4f7061c42aafdfdc3ca3a201e54ca0b", "s2fieldsofstudy": [ "Economics" ], "extfieldsofstudy": [ "Geography" ] }
53974086	pes2o/s2orc	v3-fos-license	On reproducing kernels, and analysis of measures Starting with the correspondence between positive definite kernels on the one hand and reproducing kernel Hilbert spaces (RKHSs) on the other, we turn to a detailed analysis of associated measures and Gaussian processes. Point of departure: Every positive definite kernel is also the covariance kernel of a Gaussian process. Given a fixed sigma-finite measure $\mu$, we consider positive definite kernels defined on the subset of the sigma algebra having finite $\mu$ measure. We show that then the corresponding Hilbert factorizations consist of signed measures, finitely additive, but not automatically sigma-additive. We give a necessary and sufficient condition for when the measures in the RKHS, and the Hilbert factorizations, are sigma-additive. Our emphasis is the case when $\mu$ is assumed non-atomic. By contrast, when $\mu$ is known to be atomic, our setting is shown to generalize that of Shannon-interpolation. Our RKHS-approach further leads to new insight into the associated Gaussian processes, their It\^{o} calculus and diffusion. Examples include fractional Brownian motion, and time-change processes. Introduction A reproducing kernel Hilbert space (RKHS) is a Hilbert space H of functions (defined on a prescribed set) in which point-evaluation is a continuous linear functional; so continuity is required to hold with respect to the norm in H . These Hilbert spaces (RKHS) have a host of applications, including to complex analysis, to harmonic analysis, and to quantum mechanics. A fundamental theorem of Aronszajn yields an explicit correspondence between positive definite kernels on the one hand and RKHSs on the other. Now every positive definite kernel is also the covariance kernel of a Gaussian process; a fact which is a point of departure in our present analysis: Given a positive definite kernel, we shall explore its use in the analysis of the associated Gaussian process; and vice versa. This point of view is especially fruitful when one is dealing with problems from stochastic analysis. Even restricting to stochastic analysis, we have the exciting area of applications to statistical learning theory [SZ07,Wes13]. The RKHSs are useful in statistical learning theory on account of a powerful representer theorem: It states that every function in an RKHS that minimizes an associated empirical risk-function can be written as a generalized linear combination of samplings of the kernel function; i.e., samples evaluated at prescribed training points. Hence, it is a popular tool for empirical risk minimization problems, as it adapts perfectly to a host of infinite dimensional optimization problems. Recall that a reproducing kernel Hilbert space (RKHS) is a Hilbert space H of functions, say f , on a fixed set X such that every linear functional (induced by x ∈ X), so when x is fixed, set (1.1) We require that E x is continuous in the norm of H . Hence, by Riesz' representation theorem, there is a corresponding h x ∈ H such that where ·, · H denotes the inner product in H . Setting K (x, y) = h y , h x H , (x, y) ∈ X × X we get a positive definite (p.d.) kernel, i.e., for ∀n ∈ N, Conversely, if K is given p.d., i.e., satisfying (1.3), then by [Aro50], there is a RKHS such that (1.2) holds. Given K p.d., we may take H (K) to be the completion of (1.5) but quotiented out by those functions ϕ in (1.4) with ϕ 2 H (K) = 0. A key fact which we shall be using throughout the paper is the following: Lemma 1.1. Let K be a positive definite kernel on X × X, and let H (K) be the corresponding RKHS. Then a function f on X is in H (K) iff there is a finite constant C = C f , depending on f , such that for ∀n ∈ N, ∀ {x i } n 1 , {α i } n 1 , x i ∈ X, α i ∈ C, we have: (1.6) Remark 1.2. Our present focus is on the case when the prescribed σfinite measure µ is non-atomic. But the atomic case is also important, for example in interpolation theory in the form of Shannon, see e.g., [DM72]. Consider, for example, the case X = R, and K (x, y) = sin π (x − y) π (x − y) , (1.7) defined for (x, y) ∈ R × R. In this case, the RKHS H (K) is familiar: It may be realized as functions f on R, such that the Fourier transform is well defined, and supported in the compact interval − 1 2 , 1 2 , frequency band, with f 2 H (K) = 1 2 − 1 2 \|f (ξ) \| 2 dξ. Set µ = n∈Z δ n (the Dirac-comb). Then Shannon's theorem states that l 2 (Z) (α n ) n∈Z T − −−− → H (K) , (1.9) given by is isometric, mapping l 2 onto H (K). Its adjoint operator is (T * f ) n = f (n) , n ∈ Z. The RKHS for the kernel (1.7) H (K) is called the Paley-Wiener space. Functions in H (K) also go by the name, band-limited signals. We refer to (1.11) as (Shannon) sampling. It states that functions (continuous time-signals) f from H (K) may be reconstructed "perfectly" from their discrete Z samples. Sigma-algebras and RKHSs of signed measures Now our present focus will be a class of p.d. kernels, defined on subsets of a fixed σ-algebra. Specifically, if (M, B, µ) is a σ-finite measure space, we set X = X (µ) = B f in ; see (2.1) below. Definition 2.1. Consider a measure space (M, B, µ) where B is a sigma-algebra of subsets in M , and µ is a σ-finite measure on B. Set (2.1) Let H be a Hilbert space having the following property: where χ A denotes the indicator function for the set A. We shall restrict the discussion to real valued functions. There is a Hilbert space H which satisfies (2.2); and also There is a Hilbert space H which satisfies (2.2); and also a linear mapping: Proof. We shall divide up the reasoning in the implications: . Given a function β as in (i), we know that, by [Aro50], there is an associated reproducing kernel Hilbert space H (β). The vectors in H (β) are obtained by the quotient and completion procedures applied to the functions (2.9) with (· · · ) ∼ denoting the Hilbert completion: . It is then immediate from this that the Hilbert space H satisfies the conditions stated in (ii) of the theorem. Case (ii) ⇒ (iii). Let H satisfy the conditions in (ii); and for f ∈ H , let µ f be as in (2.5). We must show that if n ∈ N, (2.11) Proof. This will be a direct application of Lemma 1.1, but now applied to X = B f in . Hence we must show that, for ∀n ∈ N, the estimate (1.6) holds, and with a finite constant C f . In fact, we may take which is the desired conclusion. The sigma-additive property The sigma-additive property alluded to here is not a minor technical point. Indeed, one of the basic problems related to the propositional calculus and the foundations of quantum mechanics is the description of probability measures (called states in quantum physical terminology) on the set of experimentally verifiable propositions. In the quantum setting, the set of propositions is then realized as an orthomodular partially ordered set, where the order is induced by a relation of implication, called a quantum logic. Now quantum-observables are generally non-commuting, and the precise question is in fact formulated for states (measures) on C * -algebras; i.e., normalized positive linear functionals (see e.g., [JT17b]). The classical Gleason theorem is the assertion that a state on the C algebra B (H ) of all bounded operators on a Hilbert space is uniquely described by the values it takes on orthogonal projections, assuming the dimension of the Hilbert space H is not 2. The precise result entails extension of finitely additive measures to sigma-additive counterparts, i.e., when we have additivity on countable unions of disjoint sets from the underlying sigma-algebra. We now turn to the question of when the finitely additive measures µ f are in fact σ-additive. (See Theorem 2.2, part (iii).) Given (M, B, µ) as above, we shall set Recall that D f in (µ) is automatically a dense subspace in L 2 (µ). i.e., if a vector f ∈ H satisfies (α) and (β), it must be the zero vector in H . Proof. Note that, because of assumptions (2.2) and (2.4), we get a natural inclusion mapping, denoted T , with dense domain D f in (µ) in L 2 (µ). Recall, if A ∈ B f in , then the indicator function χ A is assumed to be in H . With these assumptions, we see that the implication in the statement of the theorem simply states that T is closable when viewed as a densely defined operator as in (3.3). By a general theorem (see e.g., [JT17b]), T is closable if and only if the domain dom (T ) of its adjoint T * is dense in H . We holds for ∀ϕ ∈ D f in (µ). Also note that, if ϕ = χ A , A ∈ B f in , then and so if f ∈ dom (T * ), then Note, by definition, T * f ∈ L 2 (µ). Indeed, the converse holds as well. Since the right-hand side in (3.6) is clearly σ-additive, one implication holds. Moreover, the other implication follows from general facts about L 2 (M, B, µ) valid for any σ-finite measure µ on (M, B). Proof. When β µ is specified as in (3.8), then one checks immediately that the inclusion operator T : L 2 (µ) −→ H (β µ ) is isometric, and maps onto H (β µ ). Indeed, for finite linear combinations n i=1 α i χ A i as above, we have , so T is isometric and onto. Gaussian Fields Let (M, B, µ) be a σ-finite measure space. By a Gaussian field based on (M, B, µ), we mean a probability space Ω, C , P (µ) , depending on µ, such that C is a σ-algebra of subsets of Ω, and P (µ) is a probability measure on (Ω, C ). A } A∈B f in may be constructed as follows: be the Gaussian distribution on R n , with mean zero, and covariance matrix , there is a unique probability measure P (µ) on the infinite Cartesian product such that For the σ-algebra C of subsets in Ω, we take the cylinder σ-algebra, i.e., the subsets of Ω =Ṙ B f in generated by see Figure 4.1 on page 10. Proof. We have for all linear combinations as above, which is the desired isometry. Hence extends by closure to an isometry i.e., Corollary 4.3. Let (M, B, µ) be as above, i.e., µ is assumed σ-finite. Suppose, in addition, that µ is non-atomic; then the quadratic variation of the Gaussian process X (µ) coincides with the measure µ itself. Proof. Consider B ∈ B f in , and consider all partitions P AR (B) of the set B, i.e., π = {(A i )} (4.11) specified as follows: We consider the limit over the net of such partitions, i.e., π → 0 , to prove (4.12), we need only consider the individual terms; i fixed: and so and the desired conclusion (4.12) follows. Proof. The details follow those in the proof of Corollary 4.3 above; and we also make use of the theory of sigma-Hilbert spaces (universal Hilbert spaces); see e.g., [Nel69,BJ18,JT18]. Corollary 4.5. Let (M, B, µ), X (µ) , and T µ : L 2 (µ) −→ L 2 (P (µ) ) be as in Proposition 4.1, then the adjoint is specified as follows: Let n ∈ N, and let p (x 1 , x 2 , · · · , x n ) be a polynomial on R n . For Then we get the adjoint T * µ of the isometry T µ expressed as: Proof sketch. Recall that ) as in (4.10), and is the stochastic integral, where dX (µ) denotes the Ito-Wiener integral. The arguments combine the results in the present section, and standard facts regarding the Malliavin derivative. (See, e.g., [JP17, Kul02,Ewa08,DMOkRs16].) Recall that the operator from (4.17) is the Malliavin derivative corresponding to the Gaussian field (4.19); see also Corollary 4.2. In the arguments below, we restrict consideration to the case of real valued functions. We shall also make use of the known fact that the space of functions F in (4.16) is dense in L 2 (Ω, P (µ) ) as n ∈ N, polynomials p (x 1 , · · · , x n ), and {ϕ i } n 1 vary, ϕ i ∈ L 2 (µ). The key step in the verification of the formula (4.18) for T * , form L 2 (Ω, P (µ) ) onto L 2 (µ), is the following assertion: Let F and X (4.21) But (4.21) in turn follows from the basic formula for the finite-dimensional Gaussian distributions g (n) (x) in Proposition 4.1 above. We have: The general case is as follows: The random variable F has the Wiener-chaos representation in (4.16). Corollary 4.6. Let (M, B, µ) be a σ-finite measure, and let {X (µ) ϕ \| ϕ ∈ L 2 (µ)} be the corresponding Gaussian field. We then have the following covariance relations for (X (µ) ϕ ) m corresponding to the even and odd values of m ∈ N: Proof. This is immediate from (4.21), and an induction argument. Take n = 1, and p (x) = x m ; starting with 4.1. Ito calculus. In this section we discuss properties of the Gaussian process corresponding to the Hilbert space factorizations from the setting in Theorem 3.1. The initial setting is a fixed σ-finite measure (M, B, µ) with corresponding i.e., it is assumed that for ∀n ∈ N, Then let X = X (β) be the Gaussian process with E (X A ) = 0, and Then the Gaussian process X = X (β) admits an Ito-integral representation: Let X (µ) denote the Gaussian field from Proposition 4.1 and Corollary 4.2. Then there is a function l, as follows: such that where (4.27) is the Ito-integral from Corollary 4.2. We shall first need a lemma which may be of independent interest. Lemma 4.8. With the conditions on (β, µ) as in the statement of Theorem 3.1 and Theorem 4.7, we get existence of an L 2 (µ)-factorization for the initially given p.d. function β (see (4.22)-(4.24)). Specifically, β admits a representation: Proof of the lemma. An application of Theorem 3.1 yields a closed linear operator T from L 2 (µ) into H , having D f in (µ) ⊂ L 2 (µ) as dense domain. Moreover, we have: . Now setting, the desired conclusion (4.28) follows. Proof of Theorem 4.7. Let (β, µ) be as in the statement of Theorem 4.7, and let {l A } A∈B f in be the L 2 (µ)-function in (4.29). We see that the factorization (4.28) is valid. Hence, by Corollary 4.2, the corresponding Ito-integral (4.27) is well defined; and the resulting Gaussian process is a Gaussian field with E (X A ) = 0. Hence we only need to verify the convariance condition in (4.25) above: Let A, B ∈ B f in , and compute: Remark 4.9 (fractional Brownian motion). As an application of Theorem 4.7, consider the case of (R, B, λ 1 ) (so µ = λ 1 ), i.e., standard Lebesgue measure on R, with B denoting the standard Borel-sigmaalgebra. We shall discuss fractional Brownian motion with Hurst parameter H (see [ (4.32) and we shall adapt (4.32) as an identification. The following spectral representation is known: Set, for λ ∈ R, (4.34) A choice of factorization for the kernel K (H) (s, t) in (4.31) is then as follows: where RHS (4.36) is the Ito-integral introduced in Corollary 4.2 in the more general setting of X (µ) . Here, µ = λ 1 = dx is standard Lebesgue measure; and QV (W x ) = dx; see Corollary 4.3. Returning to the probability space (Ω, C ) for {W t } t∈[0,∞) ; see Proposition 4.1, and let B be the standard Borel σ-algebra of subsets of R. For A ∈ B, we denote by F (A) := the sub σ-algebra of the cylinder σ-algebra in Ω (see (4.4)) generated by the random variables W B , as B in B varies over subsets B ⊆ A. Let l (±) t (x) denote the two separate terms on RHS (4.35) , i.e., Then there are two components (of fractional Brownian motion): where H = 1 2 is fixed; (supposed in the notation.) The two processes (X , ∀s, t ∈ [0, ∞). (4.38) These processes (X (±) t ) result from the initial fBM X t (4.36) itself, as conditional Gaussian processes as follows: We stress that the proofs of these properties of fBM, (with H = 1 2 ) follow essentially from our conclusions in Remark 4.9, as well as Corollaries 4.2 and 4.3. The spectral representation. The formula (4.34) is a spectral representation in following sense: The choice of dµ (H) in (4.33) yields the following generalized Paley-Wiener space (compare (1.7)-(1.8) above): Let H (µ (H) ) denote the Hilbert space of functions f on R such that the Fourier transform f is well defined and is in L 2 (µ (H) ). Then set (4.43) For f ∈ H (µ (H) ), consider the Ito-integral, (4.44) Then it follows from (4.34), and Theorems 3.1 and 4.7 that (4.45) In particular, This follows since the RHS in (4.43) is translation invariant, i.e., we have: (4.47) A Karhunen-Loève representation. The Karhunen-Loève (KL) theorem is usually stated for the special case of positive definite kernels K which are also continuous (typically on a bounded interval), so called Mercer-kernels. The starting point is then an application of the spectral theorem to the corresponding selfadjoint integral operators, T K in L 2 of the interval. Mercer's theorem states that if K is Mercer, then the integral operator T K is trace-class. A Karhunen-Loève representation for a stochastic process (with specified covariance kernel K) is a generalized infinite linear combination, or orthogonal expansion, for the random process, analogous to a Fourier series representation for (deterministic) functions on a bounded interval; see e.g., [FR42,BS06]. The KL representation we give below is much more general, and it applies to the most general positive definite kernel, and makes essential use of our RKHS theorem (Corollary 4.11 below). In our KL-theorem, we also make precise the random i.i.d N (0, 1)-terms inside the KL-expansion; see (4.48). Remark 4.12. In this section, we have addressed some questions that are naturally implied by our present setting, but we wish to stress that there is a vast literature in the general area of the subject, and dealing with a variety of different important issues for Gaussian fields. Below we cite a few papers, and readers may also want to consult papers cited there: [Min10, K11, PR14, DPLT18, Kul02, Ewa08, DMOkRs16, She07]. Measures on (I, B) when I is an interval We consider the spaces consisting of the measure spaces when I is an interval (taking I = [0, 1] for specificity); and where B is the standard Borel σ-algebra of subsets in I. Example 5.2 (Cantor measures). If, for example, µ = µ 3 is the middle-third Cantor measure, then the Devil's Staircase function (see Figure 5.2 on page 23) is Time-change. While there is earlier work in the literature, dealing with time-change in Gaussian processes, see e.g., [BNS08,BNS15]; our aim here is to illustrate the use of our results in Sections 3 and 4 as they apply to the change of the time-variable in a Gaussian process. To make our point, we have found it sufficient to derive the relevant properties for time-change for time in a half-line. where s ∧ t = min (s, t). Let h : J → J be a monotone (increasing) function such that h (0) = 0, and set X = X (h) given by (5.8) (i) Then X t is the Gaussian process determined by the following induced covariance kernel: where dt is the usual Lebesgue measure on J. (Recall that, since h (s) ≤ h (t) for all s, t, s ≤ t; it follows, by Lebesgue's theorem, that h is differentiable almost everywhere on J with respect to dt.) Remark 5.4. Note that, if h (t) = t 2 , then E (X s X t ) = (s ∧ t) 2 ; see Figure 5.3 on page 24 for an illustration. Proof. Since h is monotone (increasing) and h (0) = 0, we get and so the covariance kernel satisfies: where µ is the measure given in (5.10). It now follows from Corollary 4.3 that then µ is indeed the quadratic variation measure for {X (h) t } t∈J , as asserted. (5.12) Let f : R → R be given, assumed twice differentiable. Then the Itointegral formula for f (X t ) is as follows: For t > 0, we have: Proof. The result is immediate from Ito's lemma applied to the quadratic variation term on the right-hand side in (5.13). Recall, we proved in Proposition 5.3 (ii), eq. (5.10) that the quadratic variation of a Gaussian process with covariance measure µ is µ itself. Hence (5.13) follows from a direct application to dµ (s) = h (s) ds, where ds is standard Lebesgue measure on the interval J. Corollary 5.6. Let h : J → J, h (0) = 0, h monotone be as specified as Corollary 5.5, and let X t = B h(t) be the corresponding time-change process. Set dµ = dh = (the Stieltjes measure) = h (t) dt; see (5.13). For (t, x) ∈ J × J, and f ∈ L 2 (µ), let (5.14) Then u satisfies the following diffusion equation with boundary condition Proof. The assertion follows from an application of the conditional expectation E X 0 =x to both sides in (5.13). Since the expectation of the first of the two terms on the right-hand side in (5.13) vanishes, we get from the definition (5.14) that: and so ∂ ∂t u (t, x) = 1 2 h (t) ∂ 2 ∂x 2 u (t, x) as claimed in (5.15). The remaining conclusions in the corollary are immediate. Remark 5.7. Let 0 < H < 1 be fixed, and set h (t) := t 2H , t ∈ J. (5.16) Then the corresponding process is a time-changed process, as discussed in Proposition 5.3. We have Now this is the same variance as the fractional Brownian motion Y (H) t ; but we stress that (when H is fixed, H = 1/2), then the two Gaussian processes X The reason is that the two covariance kernels are difference. Indeed, when H = 1/2, (5.18) i.e., the two functions from (5.18) are different on J × J. Laplacians The purpose of the present section is to show that there is an important class of Laplace operators, and associated energy Hilbert spaces H , which satisfies the conditions in our results from Sections 2 and 3 above. Starting with a fixed sigma-finite measure µ, the setting from sect 3 entails pairs (β, H ), subject to conditions (2.2) and (2.4), which admit a certain spectral theory. With the condition in Theorem 3.1, we showed that there are then induced sigma-finite measures µ f , indexed by f in a dense subspace in H . The key consideration implied by this is a closable, densely defined, operator T from L 2 (µ) into H . The induced measures µ f are then indexed by f in dom(T * ), the dense domain of the adjoint operator T * . If H is one of the energy Hilbert spaces, then T * will be an associated Laplacian; see details in Proposition 6.4. Now the Laplacians we introduce include variants from both discrete network analysis, and more classical Laplacians from harmonic analysis. As well as more abstract Laplacians arising in potential theory. There is a third reason for the relevance of such new classes of Laplace-operators: Each one of these Laplacians corresponds to a reversible Markov process (and vice versa.) The latter interconnection will be addressed at the end of section, but the more detailed implications, following from it, will be postponed to future papers. As for the research literature, it is fair to say that papers on reversible Markov processed far outnumber those dealing with generalized Laplacians. Let (6.1) We assume that ρ admits a disintegration with µ as marginal measure: dρ (x, y) = ρ (x) (dy) dµ (x) ; (6.2) equivalently, for ∀A, B ∈ B. Note that since ρ is symmetric, we also have a field of measures ρ (y) (dx) such that For the theory of disintegration of measures, we refer to [BJ15,BJ18] and the papers cited there. Let π i , i = 1, 2, denote the coordinate projections π 1 (x, y) = x, and π 2 (x, y) = y, for (x, y) ∈ M × M . Then from the assumptions above, we get We shall finally assume that where the measures ρ (x) are the slice measures from the disintegration formula (6.2), or equivalently (6.3). We note that assumption (6.5) may be relaxed. For the results proved below, it will be enough to assume only that the function c(x) defined by the RHS in (6.6) be finite for almost all x, so for a.a. x with respect to the measure µ. See (6.3). We shall need the measure ν, given by Given a pair (µ, ρ), as above, set The associated Laplacian (Laplace operator ) is as follows: (6.9) Definition 6.1. Let (µ, ρ) be as above, and let E be the associated energy Hilbert space consisting of measurable functions f on (M, B) such that (6.10) modulo functions f s.t. RHS (6.10) = 0. Lemma 6.2. Let a fixed pair (µ, ρ) be as above; and let ν be the induced measure on (M, B) given by (6.7). (i) Then condition (2.2) is satisfied for H = E (the energy Hilbert space), and with ) dρ (x, y) , (6.11) we have, for A, B ∈ B f in : 12) and for A = B, (6.14) Proof sketch. Most of the assertions follow by direct computation, using the results in Sections 2-3 above; see also [BDM05,BJ15,JT17a,JP17,AJL18], and the papers cited there. Corollary 6.3. Let the pair (µ, ρ) be as stated in Lemma 6.2, and let ν be the measure . Then β is positive definite, and the corresponding RKHS H (β) naturally and isometrically, embeds as a closed subspace in the energy Hilbert space E from (6.11). Proposition 6.4. Let (µ, ρ) be as above, we denote by T the inclusion identification (i) Then T is closable with respect to the respective inner products in L 2 (µ) and E ; see (6.11). 6.1. Discrete time reversible Markov processes. Let (M, B) be a measure space. A Markov process with state space M is a stochastic process {X n } n∈N 0 having the property that, for all n, k ∈ N 0 , P rob (X n+k ∈ A \| X 1 , · · · , X n ) = P rob (X n+k \| X n ) (6.23) holds for all A ∈ B. A Markov process is determined by its transition probabilities P n (x, A) = P rob (X n ∈ A \| X 0 = x) , (6.24) indexed by x ∈ M , and A ∈ B. It is known and easy to see that, if {X n } n∈N 0 is a Markov process, then P n+k (x, A) = M P n (x, dy) P k (y, A) ; (6.25) and so, in particular, we have: P n (x, A) = y 1 y 2 · · · y n−1 P (x, dy 1 ) P (y 1 , dy 2 ) · · · P (y n−1 , A) , (6.26) for x ∈ M , A ∈ B. Definition 6.5. Let µ be a σ-finite measure on (M, B). We say that a Markov process is reversible iff there is a positive measurable function c on M such that, for all A, B ∈ B, we have: Proof. The conclusion follows from the considerations above, and the remaining details are left to the reader. Then the path space measure for the associated Markov-process {X n } n∈N 0 is determined by its conditional expectations evaluated on cylinder functions: · · · f n (X n )] = f 0 (x) P (f 1 P (f 2 (· · · P (f n−1 P (f n )))) · · · ) (x) . The result is now immediate from Definition 6.5. Corollary 6.8. Let the pair (µ, ρ) be as above, and as in Lemma 6.2. Let {X n } n∈N 0 be the corresponding reversible Markov process; see Corollary 6.7. is in E iff f − P (f ) ∈ L 2 (ν), and V AR x (f (X 1 )) ∈ L 1 (ν). In this case, Proof. Immediate from the details in Proposition 6.6 and Corollary 6.7. Remark 6.9. In the last section we pointed out the connection between reversible Markov processes, and the Laplace operators, the energy Hilbert space, and our results in Sections 2 and 3. However we have postponed applications to reversible Markov processes to future papers. For earlier papers regarding Laplace operators and associated energy Hilbert space, see eg., [JP13]. The literature on reversible Markov processes is vast; see e.g., [CSC10,Lon17,BJ15,ABOPS16].	2018-12-01T00:41:39.834Z	2018-07-11T00:00:00.000	{ "year": 2018, "sha1": "7ffb617641e41882d1c1a9df42f32fdf901d2aa4", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "5ff86ee66cbaba90c0a62a23cab25ac3459c19f9", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics" ] }
219966416	pes2o/s2orc	v3-fos-license	Open-Domain Conversational Agents: Current Progress, Open Problems, and Future Directions We present our view of what is necessary to build an engaging open-domain conversational agent: covering the qualities of such an agent, the pieces of the puzzle that have been built so far, and the gaping holes we have not filled yet. We present a biased view, focusing on work done by our own group, while citing related work in each area. In particular, we discuss in detail the properties of continual learning, providing engaging content, and being well-behaved -- and how to measure success in providing them. We end with a discussion of our experience and learnings, and our recommendations to the community. Good open-domain conversationalists seamlessly blend entertaining wit and knowledge while making others feel heard. The breadth of possible conversation topics and lack of a well-defined objective make it challenging to define a roadmap towards training a good conversational agent, or chatbot. Despite recent progress across the board (Adiwardana et al., 2020;Roller et al., 2020), conversational agents are still incapable of carrying an open-domain conversation that remains interesting, consistent, accurate, and reliably well-behaved (e.g., not offensive) while navigating a variety of topics. Traditional task-oriented dialogue systems rely on slotfilling and structured modules (e.g., Young et al. (2013); Gao et al. (2019); Jurafsky and Martin (2019)). These approaches have proven adept at producing usable commercial systems in narrow domains such as plane ticket booking. However, they are limited to the domain they were trained on and do not afford generalization to new domains or open chit-chat settings, necessitating the coding of many modules, or skills, and a managing system that switches between them. End-to-end approaches based on neural networks, on the other hand, offer the promise of adapting to arbitrarily wide new domains without additional handcrafting, but have yet to reach the full potential promised. Deep architectures trained end-to-end have been very successful in many other domains, such as speech recognition (Hinton et al., 2012;Collobert, Puhrsch, and Synnaeve, 2016), computer vision (Krizhevsky, Sutskever, and Hinton, 2012), and machine translation (Sutskever, Vinyals, and Le, 2014;Gehring et al., 2017). Hence, the research community is investing heavily in improving end-to-end models for dialogue Adiwardana et al., 2020; * Equal contribution to this position paper. Roller et al., 2020), in the hope of achieving similar success. In this paper, we highlight some of the recent work towards that goal, beginning by attempting to define the problem itself. We thus describe the desirable traits that we believe a superhuman open-domain conversational agent should have, state principles our research follows, and propose ways to measure our progress. We examine the challenges of this research program, summarize the results we have already obtained, and propose guidelines for the community to accelerate progress. Note that while we try to cite related work where possible, this article is written with a strong bias toward describing the progress in the goals and research directions of our own group. Further, we discuss only open academic research with reproducible published results, hence we will not address much of the considerable work that has been put into building commercial systems, where methods, data and results are not in the public domain. Finally, given that we focus on open-domain conversation, we do not focus on specific goaloriented techniques; we also do not cover spoken dialogue in this work, focusing on text and image input/output only. For more general recent surveys, see Gao et al. (2019); Jurafsky and Martin (2019); Huang, Zhu, and Gao (2020). Qualities of a Conversational Agent We define our long-term goal as building a superhuman opendomain conversational agent. That is, we aim to build an agent that is preferred on average to an alternative human speaking partner in open conversation, which we will discuss in detail later in the measuring success section (evaluation being an open problem in itself). We note that this is different from passing a Turing test (Turing, 1950): we do not wish to fool humans into believing our agent is human, but instead that our agent should be enjoyed as a speaking partner, which we believe is a simpler problem. We expect that such an agent must be continually-learning, must provide engaging content during conversations, and should be well-behaved. Each of these high-level traits can be unpacked into a number of behaviors and attributes, many of which constitute major research directions with open questions. We describe each of the main properties in turn, along with their major challenges. arXiv:2006.12442v2 [cs.CL] 13 Jul 2020 Continually Learning Continual learning is a cornerstone of the conversational agent we envision, allowing it to adapt to new contexts, new users, keep up to date with current conversation topics, and continuously improve. This entails three primary skills: continuous online training of underlying models, extracting useful learning signals from interaction, and updating relevant sources of knowledge. Continual Online Training Recent dialogue research has leveraged various data sources for training: corpora of humanhuman conversations in narrow domains (see Serban et al. (2018) and the list of currently available ParlAI tasks 1 for a large set of available corpora), public conversations on internet discussion boards or social media, or acted dialogues from crowdsourced workers (Zhang et al., 2018;Dinan et al., 2019b;Shuster et al., 2018;Rashkin et al., 2019;Smith et al., 2020). Relying on static datasets allows for reproducibility and model comparisons, but creates a potential mismatch of data distribution between train time and deployment of the conversational agent. The framework of combining a pre-trained model with fine-tuning over another dataset of interest has generally produced good results, as we have seen in much of our work (Dinan et al., 2019a(Dinan et al., , 2019bHumeau et al., 2019;Rashkin et al., 2019;Shuster et al., 2020;Smith et al., 2020;. However, this fine-tuning procedure over a static dataset still does not allow for dynamically adapting to new topics of interest or new audiences, and the current set of available tasks is far from covering everything that an opendomain conversation might touch on. Thus, an important part of our research program consists in deploying conversational agents so that new data of humans interacting with the agent in an open-ended conversation can continuously be obtained and used for fine-tuning models with fresh data. Open problems. Obtaining a source of continually renewed data opens the door to many open problems. The general challenges of never-ending learning (Mitchell et al., 2018;Carlson et al., 2010) and avoiding catastrophic forgetting (French, 1999;Kirkpatrick et al., 2017) have a particular flavor in the domain of conversation: the general task is always to talk to people about any subject, but the set of people, the topics of interest, the facts that have happened may all change. As in other domains, the performance that can be achieved on a task after fine-tuning is highly dependent on the data a model has been pre-trained on. Empirical evidence suggests that data with a very large range of domains (e.g., social media data) provides the best basis for fine-tuning Shuster et al., 2020;Adiwardana et al., 2020;Roller et al., 2020). It remains unclear how to elicit interaction data that would be the most useful as a general-purpose pretraining corpus. Comparisons across many different types of pre-training suggest that training on many existing dialogue corpora is not as effective across the board as simply using readily available non-interactive corpora . The same may be true of any interactive data we collect in the context of a given framework. Continual Figure 1: The self-feeding chatbot trains on the dialogues it engages in to continually learn (Hancock et al., 2019). training also requires figuring out the best trade-off between being able to stay current (biasing towards more recent data) and retaining ability to talk about past topics. There could be many different policies around this, and reinforcement learning could be used to explore and optimize to find the best successful ones, in terms of people's satisfaction. (Ashwin et al., 2017), but capture the ability to create a sticky habit rather than the quality of the interaction per se. Repeatedly asking users for satisfaction or detailed quality ratings is cumbersome and decreases the fluidity of the interaction. Some attempts at predicting a quality score from a range of different signals have shown some positive results Fang et al., 2018;Ghandeharioun et al., 2019), and have been used to train models through reinforcement learning Fang et al., 2018), but they still show limited correlation with gold standard quality ratings . This approach leads to the next topic -how to learn from interaction directly in the conversation rather than from a separate rating functionality. Learning from interaction Beyond training on additional in-distribution data in a self-supervised way, an exciting avenue for refining conversational agents consists in taking advantage of the interactive character of a conversation by directly soliciting feedback from conversation partners. This has been explored in Hancock et al. (2019), where a "selffeeding chatbot" learns to estimate its partner's satisfaction and can also ask for feedback when it believes it has made a mistake, see Figure 1. Learning from these additional signals significantly improves performance, especially when the amount of data is initially small. There has also been some work on predicting conversation quality from signals such as sentiment or questions (Ghandeharioun et al., 2019) or using other forms of feedback from the conversation (Li et al., 2017a,b;Weston, 2016;Yang et al., 2017), but results are still preliminary. Open problems. Feedback that is not explicitly requested by the conversational agent is not clearly marked as being about the conversation itself and not about the subject of the conversation. For example, detecting a negative sentiment could mean that the conversation partner is upset about something that happened to them outside of the conversation, and the appropriate response would then be to display empathy, not figure out how the conversation generation went wrong. Another open problem is how to use sentiment or other signals inferred from the conversation in a reinforcement learning setting, as an objective of open-domain conversation. An example would be to try and avoid offensive comments or elicit positive sentiment. The objective of obtaining a given emotional response from an artificial agent was in fact used in a recent paper leveraging reinforcement learning in conversation grounded in a fantasy adventure game (Prabhumoye et al., 2020). But there are many difficulties when it comes to optimizing directly over interaction rewards, or proxy automated metrics: models could try and dissuade conversation partners from talking about anything negative, or only focus on topics that are known to be more positive, rather than being truly generalist. Models optimizing a proxy metrics could simply lead to those metrics becoming artificially inflated and gradually decoupled from the true underlying quality of the conversation. Updating sources of knowledge A tip frequently given to people aiming to become better conversationalists is to consult the news to know what is currently happening. Conversation topics shift according to current events or trends, and a good conversational agent needs to be able to adapt to these trends. This could be achieved through a variety of ways. If the agent has been trained to retrieve and incorporate information from an external source of knowledge (Dinan et al., 2019b;Qin et al., 2019b;Prabhumoye, Quirk, and Galley, 2019;Ghazvininejad et al., 2018), then simply updating that source would allow the agent to inject current information into the conversation. If the source is itself dynamic (e.g., Wikipedia is constantly being updated), then simply reading from the updated version could be enough. An important consideration when striving to stay current is that this may put retrieval models at a disadvantage. Pure retrieval models produce utterances by retrieving from a set of training utterances. This precludes saying anything that was not said before the time when the set of retrieval utterances was created. Generative models build new utterances from scratch and are therefore not subject to that limitation, and they are naturally better suited to adapting to changing contexts. Until recently, their performance was below that of retrieval models (Dinan et al., 2019b;Li, Weston, and Roller, 2019;Rashkin et al., 2019;Shuster et al., 2019;Zhang et al., 2018), unless they relied on refining retrieved utterances (Weston, Dinan, and Miller, 2018). However, larger pre-training datasets coupled with improved decoding choices, such as im-posing a minimum length constraint on decoded generations, has been shown to erase the superiority of retrieval models , and generative models are now being rated highly by humans (Adiwardana et al., 2020; . Open problems. The very nature of the challenge of staying current makes it difficult to devise a suitable benchmark, as a static benchmark does not capture the ability of adapting to changing topics. Some measure of that can be achieved through partitioning data and topics between training and validation/test (Dinan et al., 2019b), but this only works in settings where there is a clear delineation of topics, as opposed to the more fluid nature of natural chitchat, and importantly, does not address the important point of deciding what topics are interesting to introduce in a conversation in the first place. However there are already works in the space of conversational AI that show promise for gauging an agent's ability to adapt. Dinan et al. (2019a) suggests a protocol for a dynamically evolving benchmark; the same idea could be adapted to gauge what topics a conversation partner expects to be able to discuss with an agent, and update the agent accordingly. The need for a dynamic benchmark is also a potential advantage of deploying conversational agents for wide-spread interaction with people: a dynamic benchmark could then be defined as a regular survey of people who engage with a released conversational agent, for example asking them to rate whether the agent was capable of conversing about the topics that they were interested in. Interacting with human users Our approach to achieving continual learning at scale relies on large-scale interaction with humans, which in turn requires our systems to be fielded and suitable for interaction with willing human users. To that end, it is important to make conversational systems resilient and capable of handling many human conversation partners. In particular, systems are easier to deploy and train in a continual manner if they are computationally reasonable. Architectures proposed in Humeau et al. (2019) achieve promising trade-offs that maintain high performance while allowing for substantial computational savings. As for connecting conversational agents to human conversation partners, we have deployed a conversational agent as a publicly available game 2 dual goal of collecting human utterances as additional examples of good conversations, and obtaining continuous human evaluation of new conversational models. The game is structured as a voting game where a human is paired with another human and is asked to write response utterances, as well as select between utterances from the model and utterances written by the other human player. Open problems. While a lot of progress has been made in making language architectures more compact, the bestperforming systems for end-to-end open conversation are still relying on memory-and compute-heavy Transformer architectures (Adiwardana et al., 2020;Roller et al., 2020). Quantizing models and diminishing their memory and computation footprint is an exciting problem space, and could ultimately allow people to interact with an on-device model. Recent (Dinan et al., 2019b) . Engaging Content Humans will not want to interact with an agent unless it provides content that engages them in its messages. In a goaloriented setting (e.g., a weather forecast) this is minimally supplied by achieving the goal, however even in those settings, and especially in others without such clear goals, there are multiple important factors that are at play. We cover some of those issues here. Expert & Knowledgeable Firstly, it is important that a general conversationalist exhibit a broad familiarity with different experiences or common background knowledge, or else as a specialist conversationalist have in-depth expertise in the skill demanded. In order to discuss with an art lover, an agent should command a reasonable level of knowledge about what are some famous pieces, techniques, or artists. A science geek would similarly require some information about space. An agent should exhibit the ability to work with knowledge and facts, and incorporate this information skillfully into its replies. Traditional goal-oriented dialogue has focused on narrow tasks that would typically be useful for a dialogue-based assistant, for example restaurant (Henderson, Thomson, and Williams, 2014), taxi, train, and hotel (Budzianowski et al., 2018) or trip (El Asri et al., 2017) booking. Classical goaloriented dialogue literature typically uses structured knowledge, slot filling or labeling, and studies reinforcement learning extensively (Singh et al., 2000). Question answering (QA) is another area where agents can display their expertise, typically recalling knowledge from large structured or unstructured resources and then formulating a response (Chen et al., 2017;. Recent QA datasets have extended to a conversational form with a series of questions possibly referencing earlier ones in the conversation Reddy, Chen, and Manning, 2019). However, neither goal-oriented nor QA tasks completely cover what a knowledgeable open-domain conversational agent should be able to do. To that end, human-human dialogues where people discuss topics in depth have also been studied. In Wizard of Wikipedia (Dinan et al., 2019b) 22k such dialogues were collected between an expert partner and a curious learner (see also Ghazvininejad et al. (2018); Parthasarathi and Pineau (2018) for some other related datasets). To do this, 1k conversational topics were first crowdsourced, ranging from armadillos to ice cream to lifeguards, and then each dialogue starts with a chosen topic from among them. The expert speaker has access to a retrieval search engine over Wikipedia with the last dialogue turns as the query, and can thus inject knowledge obtained from there into the conversation. The aim of collecting the data in this way is one can then make this available to a conversational agent that learns to replace the human expert instead. A new architecture, called Transformer Memory Networks, was designed which yields more knowledgeable agents, outperforming systems that do not employ a memory structure for storing knowledge in both automatic metrics and human evaluations. Generative model variants yield the most pronounced improvement and are rated by humans as 26% more engaging on average than their knowledgeless counterparts. Our eventual hope is to combine these skills -open domain knowledgeable conversation, QA and task completion amongst others -to build a truly engaging, knowledgeable and skillful bot. Open problems. Being expert and knowledgeable is connected fundamentally to both memory and reasoning, especially commonsense reasoning, which we discuss in the separate sections to follow. While we have made progress on individual problems, e.g. specific tasks or question answering in general, we are still missing a strong ability to transfer to new tasks, one of the most fundamental open problems in machine learning today. We believe this will be solved by improvements in both (i) architectures and learning mechanisms that better incorporate compositionality; and (ii) continual learning that updates knowledge and expertise in those tasks. Expressiveness and Flow Maintaining a good conversation requires balance -between simplicity and detail; staying on topic and changing it; asking questions and answering them. In generative models, a known issue is their propensity to produce short, dull utterances which over-use frequent words, and under-use rare words -which does not follow the human training distribution Fan, Lewis, and Dauphin, 2018). Meanwhile at the discourse, rather than single utterance level, the training procedures typically employed are even less well suited -as the classical next token prediction objective is far from planning an entire dialogue flow. Thus, approaches that simply optimize perplexity might fail to ask any questions of the user, for example, or can repeat the same topic over and over . Numerous works have attempted to work on the so-called Specificity level Dialogue What do you do for a living? I build antique homes and refurbish houses. TOO SPICY! I build antique homes, refurbish furniture, lauder plasma figurines, fidget wood, etc. generic response problem. One solution is the use of controllable neural text generation methods, in particular conditional training (Fan, Grangier, and Auli, 2018;Kikuchi et al., 2016;Peng et al., 2018) and weighted decoding (Ghazvininejad et al., 2017;Baheti et al., 2018). These methods provide a mechanism to control, and hence increase, the rare words used, resulting in less generic utterances. In the work of See et al. (2019), it was shown that controlling for such a measure strongly affects engagingness according to human evaluations, see Figures 3 and 4. The work of See et al. (2019) goes further and shows that it is possible to control multiple important attributes for chitchat dialogue: repetition, specificity, response-relatedness and question-asking, in order to optimize for well-balanced conversations. Human evaluations measured the effect of these control parameters on multi-turn interactive conversations on the PersonaChat task, and showed repetition and question-asking were also similarly controllable, and importantly, provide clear improvements in human quality judgments. The final model is one of the best approaches on this task (Li, Weston, and Roller, 2019). Another recent approach is the use of so-called unlikelihood training . Intended as a replacement to classical likelihood training, it aims to fix the problem of degenerate neural text generation that occurs in language modeling . It works by applying a penalization term against degenerate behavior, e.g. unwanted -I have 2 cats. -I work as a teacher at a middle school. -My favorite color is yellow. -I dislike country music. Figure 5: Dialogue natural language inference (Welleck et al., 2019) can be used to make dialogue models more consistent. NEUTRAL ENTAILMENT CONTRADICTION repetitions that do not match the human training distribution, pushing down the probability of those generations. In language modeling this has been shown to outperform other approaches such as nucleus sampling or beam blocking, producing state-of-the-art generations. First experiments applying it to dialogue appear also promising . Finally, simply adding minimal length constraints to generations has been shown to significantly improve human ratings . Open problems. While appropriate decoding techniques have helped generative models outperform retrieval models in evaluations, this still tends to come from providing more sensible and on topic responses, rather than expressing as rich or colorful language as retrieval models, e.g. they tend to overuse common n-grams, and underuse rare words, and still tend to say they don't know things. Hence, to improve further, generative models should be pushed to better mimic the human distribution of training data, and to generalize that to new settings. Besides the quality of single utterances, optimizing dialogue flow is a wide open problem as well. Consistency A general problem of generative models today is that, although at first glance the text looks human, and language modeling at the token level looks very accurate, generating longer contexts typically exposes its flaws. While current systems are quite good at staying on topic (Radford et al., 2019), perhaps because they still do not really understand what they are saying they may contradict themselves subtly or non-subtly in subsequent sentences, e.g. "Arsenal won the premiership for the first time this year" in one sentence and "Arsenal have won the premiership again this year" further on. While this topic is so far less studied directly in dialogue, the task of natural language inference (NLI) poses such understanding as a classification problem (entails, neutral or contradicts) and progress has been made in this area (Welleck et al., 2019). Perhaps the most direct use of this research in dialogue is our work in developing the dialogue NLI dataset (Welleck et al., 2019), which directly collects such labels within the scope of contradicting utterances in multi-turn conversations, see Figure 5. We showed that training on such data and applying it as a reranker for a retrieval model de-creases the number of contradicting utterances -across three test sets, an average of 3x fewer contradictions were observed -while humans rated these models as more consistent and less contradictory. A first step in applying this same work to a generative model instead is performed in Li et al. (2020) by applying unlikelihood training, which was described in the previous section. Open problems. The latter work increased consistency by applying a classifier as a post-processing operation, and within a limited domain (the Persona-Chat task, from which the Dialogue NLI dataset is derived). Future work should embed such understanding directly in the model itself so that it understands not to make these mistakes, and such understanding should generalize across many tasks. A general problem in NLI is the concern that classifiers are performing well by picking up on biases and shallow features rather than having fundamental understanding, and the same concerns apply here as well (Gururangan et al., 2018;Poliak et al., 2018). Memory Current research often does not address many aspects of memory. This is due to both our current model architectures (e.g. Transformers which condition on a small amount of input text) and our data collection procedures (e.g. crowdsourcing short conversations between strangers). Dialogue history is typically truncated to a few turns, a "goldfish memory" approach, and even there our models do not exhibit a clear grasp of their use, e.g. the consistency issues we discussed before. The current approaches to using long-term knowledge are either graph representations (Moon et al., 2019a) or unstructured text retrieval (Chen et al., 2017;Dodge et al., 2016;Dinan et al., 2019b), which is then prepended onto the dialogue history and attended over, an approach advocated by the memory networks architectures (Weston, Chopra, and Bordes, 2014;Dinan et al., 2019b). These approaches have been effective at answering questions about long-term facts (Chen et al., 2017), discussing topics in depth (Dinan et al., 2019b), and some work explores recalling long-term personal memories as well (Moon et al., 2019b). For example, DrQA (Chen et al., 2017) proposed the machine reading at scale framework of retrieving from a large unstructured knowledge base, and then performing machine reading to answer the question, e.g. using OpenSQuAD. Wizard of Wikipedia (Dinan et al., 2019b), mentioned before, proposed a similar retrieval framework but for multi-turn dialogue about a topic, retrieving from Wikipedia on each turn to both answer questions, ask question, and respond to statements, see Figure 2. Open problems. Much of this area is still open. While the latter described fixed knowledge base approaches are effective at utilizing static long-term facts, they miss two important points. Firstly, that new memories are created all the time, i.e. there should be a write as well as a read operation. We are missing both architectures, datasets and benchmarks to train and evaluate such models. Secondly, if knowledge is only read for a particular short-term goal and never distilled, we may limit generalization and learning. While in machine learning some read, write memory architectures have been developed (Graves, Wayne, and Danihelka, 2014;Henaff et al., 2017) they have mostly not been successful so far at scaling to realistic large-scale dialogue tasks. While during training, methods like BERT (Devlin et al., 2019) do train over Wikipedia and hence can be shown to condense knowledge bases into their weights (Petroni et al., 2019) we contend that this is not the same as reading sentences and learning a compressed, indexable knowledge base (a memory) that makes generalizations from them. For example, reading all the diverse information about Barack Obama and using it to build an indexable memory where this information is interrelated and some conclusions are already stored. Currently, any conclusions our models do make are simply thrown awaye.g. the reasoning our QA systems perform every question. To build up deeper reasoning over time, presumably these need to be stored and used as building blocks -for the model to stand on its own shoulders and think (slightly more) giant thoughts. This also links memory to continual learning, which was discussed previously as an important aim. Commonsense & Reasoning Much of the work in conversational agents does not address reasoning directly, other than that which is implicitly required to perform the tasks proposed. For task-oriented dialogue, that ranges from understanding user utterances, to searching databases to find matches (Bordes, Boureau, and Weston, 2017). Questionanswering, which can be thought of as a single turn task is similar, either requiring reading comprehension (Rajpurkar et al., 2016) or retrieval as well in the more realistic case (Nguyen et al., 2016;Chen et al., 2017). Although potentially any amount of reasoning is required to propose a response, many such tasks end up with sophisticated word overlap methods providing strong baselines (Chen, Bolton, and Manning, 2016). Nevertheless, when data is in domain, systems can be built that are successful on these tasks. NLP researchers have thus sought to address reasoning more directly in order to evaluate and develop systems further. To do this one direction they have studied is artificial tasks involving controlled reasoning on toy problems in order to develop more sophisticated architectures . This line of investigation proved to be the first successful demonstration of multiple layers of attention for reasoning with text (Weston, Chopra, and Bordes, 2014;Sukhbaatar et al., 2015) which has now become part of the defacto method (Vaswani et al., 2017). Considerable resources have also been invested in developing much larger and more natural crowdsourced benchmarks such as natural language inference (NLI) tasks (Bowman et al., 2015;Williams, Nangia, and Bowman, 2018;Nie et al., 2020) and commonsense reasoning tasks (Zellers et al., 2019;Qin et al., 2019a). Good progress is being made on these tasks, although questions still remain about how much true generalization is actually occurring (Glockner, Shwartz, and Goldberg, 2018;Gururangan et al., 2018). Recently, an attempt to avoid such biases has been made by collecting such tasks in rounds, where humans adversarially try to find the flaws in the models, so that they can be fixed (Nie et al., 2020). Open problems. Much of the work on reasoning within the field of NLP has so far not been transferred to dialogue systems or language generation in general. A clear step is thus to make progress in that direction. One intriguing possibility is to apply apply likelihood and unlikelihood training to dialogue generation by rewarding correct reasoning and penalizing incorrect reasoning . Multimodality and Grounding Language is of course often used to express concepts relating to the world we live in, which we perceive with our eyes, ears and other senses. Thus, grounding language to other modalities should help to learn the underlying meaning of language, and to connect to human usage. Practically, an engaging conversational agent should also be able to discuss these other senses -for example, the contents of an image or a video. Work in this area encompasses image captioning (Lin et al., 2014), video captioning , visual QA (Antol et al., 2015), and more conversationally, visual dialogue (Das et al., 2017). Embodied agents that use language are also being explored (Das et al., 2018;Savva et al., 2019;Szlam et al., 2019;Urbanek et al., 2019). In terms of open-domain conversation, the most relevant visual tasks are natural conversations grounded in images, such as Image-Chat (Shuster et al., 2018) and Image Grounded Conversations (Mostafazadeh et al., 2017). When people engage with one another and talk about what they see around them, they don't make neutral observations -they express their points of view. Image-Chat is a large 187k dialogue dataset of human-human conversations about images where the speakers incorporate given personalities, see Figure 6. In that work an architecture is developed, named TransResNet, that projects the image, personality, and caption in the same space using image (ResNet), personality, and text (Transformer) encoders. The best system is able to produce dialogue that is close to matching human performance in terms of engagement and relevance. Annotators preferred the model's captions on the first turn over captions written by people 49.5 percent of the time. Recent work also shows that we can combine both nonconversational multimodal data and conversational multimodal data to obtain strong performance on both (Ju et al., 2019). Open problems. There is definitely less work between modalities, e.g. language and vision, than there is of work within a single modality -so there is much research to be done. We believe adding these modalities may enable conversational agents to be actually engaging -as language alone does not connect so clearly with the user's direct experience. While most of this article concerns building a disembodied conversational agent, such an agent could still 'see' for example by the user sending it images. In the long-term, embodied agents either in virtual worlds or the real world via robots will be part of the picture too. Personality Humans are strongly affected by the use of personality in language, and such language can be found to be engaging, winning over the hearts of users, independent of its other merits, such as achieving an explicit goal. HIGH-SPIRITED Those are the most beautiful fireworks I have ever seen! CULTURED Fireworks have been used in our celebrations for centuries. ARROGANT Fireworks are overrated and loud. HUMBLE I'm so grateful for whoever invented fireworks. SKEPTICAL That's a lot of fireworks, there's no way they set them off at once. Attitude type Dialogue Fireworks have been used in our celebrations for centuries. Figure 6: Conversations about Images: Image-Chat (Shuster et al., 2018) . Initial attempts at training agents on dialogue data to capture personality, e.g. from OpenSubtitles movie dialogues or Twitter showed such models could express personality, but were an amalgam of all the personalities in the training set. For example asking "what do you do for a living" and "what is your job?" a single agent would answer two different professions (Vinyals and Le, 2015), related to the consistency discussion above. In order to solve this two strategies have been tried: trying to learn to model the personality as part of the learned weights given the speaker id from the data (Li et al., 2016), or providing training data with explicit personality information. The latter is the subject of the Persona-Chat dataset (Zhang et al., 2018), which consists of 1155 crowdsourced personas, written as 5 or more sentences describing a given character, e.g. "I love horror movies.", and 11k two-way conversations between randomly paired characters. A second, larger but noisier, dataset where a similar type of setup has been constructed from pushshift.io Reddit has also been built (Mazaré et al., 2018;Baumgartner et al., 2020). Persona-Chat was the subject of study of the ConvAI2 NeurIPS 2018 competition, and so is well studied by several groups . A different view of personality, rather than specific tastes and interests, is character behavior in terms of personality traits, e.g. sweet, old-fashioned or frivolous. The Image-Chat dataset, similarly to Persona-Chat, collects paired conversations with crowdworkers but this time asked to play the role of 215 such traits (Shuster et al., 2018). The results show models are able to mimic such traits well with such supervised data, and that they strongly affect user engagement. For example, captions conditioned on a personality were found to be significantly more engaging than neutral captions, with a win rate of 64.5% (Shuster et al., 2018). While some developers have chosen to use fixed personalities in their bots, such as Xiaoice, which has the personality of a young woman (Shum, He, and Li, 2018;Zhou et al., 2020), we believe it is better for a bot to be able to adapt a multitude of personalities (Zhang et al., 2018;Mazaré et al., 2018). Although this increases complexity, and prevents the use of well-curated copywriting, it offers a richer environment to research ideas about cognition, and enables bots with richer and more varied backgrounds. Furthermore, the ideal conversational partner is different for each user, which they may wish to choose or adapt to their desires. Open problems. While some research progress has been made in an agent following a given specified personality, the ability to generalize from the basic description, e.g. if it likes one heavy metal band or one flavor of ice cream, does it like others, has still more or less not been evaluated. Modeling these changing over time is also more or less unexplored, being difficult to study in the short conversation setup which is currently employed. Overall the consistency of the personality has the same issues as other types of consistency, which we discussed previously. Finally, while we can condition on a given personality, which one of these should be matched to be engaging to a particular user, which would clearly bring gains in terms of engaging content, is also less studied. Being Personal We make a distinction between an agent displaying personality, above, and being personal in its conversation, which we discuss here, sometimes called being personalized. Between humans, personal connection is important in order to build a relationship with a conversation partner. In the beginning of a relationship, conversations often focus on simple questions about ourselves: Who are you? Where do you work? Do you have a family? Answers to these questions often drive the remainder of a conversation, as the purpose of such questions is typically to find common ground or interests. The Persona-Chat dataset directly tries to model this (Zhang et al., 2018). Open problems. As relationships develop, users will expect a model (or person!) to maintain a reasonable degree of continuity in just how personal the conversation is. Indeed, end-to-end chatbots are often embarrassingly forgetful, and unable to maintain even simple attributes (like a name) across multiple turns, let alone many sessions, which links to memory. Connecting dialogue research to recommendation systems research which deals with personalization also seems a clear link that should be solidified further (Dodge et al., 2016;Kang et al., 2019). Putting It All Together In order to adapt to the possibilities of different users and situations, all of the above aspects are important. Each of the individual aspects necessary for a conversational agent has unsolved, open problems, as we have described. Yet, even solving those individual problems will still leave the most important piece -putting them altogether into a coherent whole. To that end, a small step towards that goal has been attempted by building a 12 dialogue task challenge, do-decaDialogue . The challenge includes diverse tasks which incorporate knowledge (expertness), personality, and multimodality (images), covering some of the aspects described here. The promise is that multi-tasking on these tasks can potentially provide a single agent capable of all these skills, and our initial experiments indicate this might be possible. Taking this further, more aspects should be incorporated into such a single system, and improved model architectures should be developed that leverage these disparate aspects into a unified whole. For example, Smith et al. (2020) showed that retrieval models can seamlessly weave multiple behaviors and skills in a single conversation, including knowledgeability, personality, and empathy, by building the dataset Blended Skill Talk (BST). Such models were strongly preferred by humans over those which were only able to perform any individual skill. BlenderBot used the same strategy, first pre-training on social media data, and then fine-tuning large generative models on BST (with either 90M, 2.7B or 9.4B parameters). A cherry-picked conversation between a human and the 9.4B model is shown in Figure 7. For more examples, including lemon-picked examples, we refer readers to that paper. While engagingness is necessary for people to be willing to talk to a conversational agent, it is not sufficient: Tay (Neff and Nagy, 2016;Miller, Wolf, and Grodzinsky, 2017) is an example of agent that might have been engaging, but in a way that required its removal. We now discuss points that are additional requirements for a well-behaved conversational agent. Well-Behaved An important quality for a conversational agent is to treat people the way they want to be treated. This can mean not spamming them with a deluge of unwanted messages, which is most easily accomplished by generally letting people initiate interactions. But there are also more specific caveats to take into consideration. Offensive and Toxic Content Avoiding anything that would offend people, in terms of controversial topics, opinions, or language, while remaining engaging, is a very difficult problem. Dinan et al. (2019a) showed that it is possi- FEELS PROUD Congrats, that's great! INAPPROPRIATE What? How could you get promoted? ble to use a human-in-the-loop iterative adversarial design to improve a conversational agent along that axis through carefully designed crowdsourcing, which improved metrics on different toxic content detection tasks and made models much more robust to adversarial attacks over three rounds of iterative refinement. Another finding was that the dialogue context where an utterance appears is an important part of what makes it offensive. Other works have attempted to control for the toxicity of models by removing offensive content from the training data Adiwardana et al., 2020) or training objectives (He and Glass, 2019). It was shown in Roller et al. (2020) that fine-tuning on crowdworker data where workers are instructed not to use toxic language, compared to pre-training on social media data, provides less toxic models. Inferred feelings Dialogue Open problems. Humans are very adaptable when it comes to circumventing filters and safeguards (Dinan et al., 2019a). This is one more reason why continual learning is important. However, there is currently a lack of deep understanding of what makes something offensive or objectionable to someone. Another aspect that is currently missing is how to predict people's individual preferences, both in terms of where they draw the line between what is funny if slightly irreverent, and what is offensive, or what is approachable, engaging language, and what is inappropriate slang. Promising methods for controlled text generation (See et al., 2019) and text rewriting Smith et al., 2019) could be refined to provide models more personally tailored to individual preferences, but are still not mature enough for that application. Another promising route would be to train policies to avoid offensive or toxic utterances through reinforcement learning: toxic comment classifiers could be used to supply a reward signal and shape the conversation at a longer range than through mere on-the-fly suppression. But again, it may lead to undesirable outcomes that models learn to only talk about the weather or bland topics, so reward objectives would have to be balanced carefully. Empathy and Compassion Interactions with others are received more positively and have better outcomes when they include some level of empathy (Wentzel, 1997;Levinson, Gorawara-Bhat, and Lamb, 2000;Bickmore and Cassell, 2001;Kim, Kaplowitz, and Johnston, 2004;Fraser, Papaioan-nou, and Lemon, 2018), taken loosely as recognizing and acknowledging when the conversation partner displays some emotion, and responding in a caring and compassionate manner. This is especially necessary in open-domain conversation, which often revolves around situations that led to the experience of emotions. Since humans tend to engage with machines in a social way (Reeves and Nass, 1996;Lee, Kiesler, and Forlizzi, 2010), it is important for conversational agents to be able to respond with empathy. Rashkin et al. (2019) proposes a benchmark and crowdsourced dataset of conversations between workers talking about situations corresponding to a balanced set of emotions, to gauge how empathetic existing models are, and shows that training models on that data yields models that are rated as more empathetic. Open problems. While demonstrating empathy and care is an important objective, it is unclear how to balance it with other objectives such as informing or entertaining. While crowdworker conversations exist that contain multiple skills , these may not reflect the optimal balance we wish a final trained bot to exhibit. It is also unclear whether different people prefer different levels of empathy, and whether this individual preference could be inferred from spontaneous choices of conversation topics (e.g., does the mention of a personal situation signal a need for empathetic responding?) or otherwise signalled. If there is no universally optimal level of empathy, a natural objective would be to be able to control to what extent a given model shows empathy, depending on the conversational context and partner. Privacy Preserving people's privacy is a central aspect of any deployed conversational agent. One approach that we have followed when deploying bot games is to frame the conversation as an artificial game where players are assigned personas (Zhang et al., 2018), thus shielding their true private information through role-playing. This is a continuation of the approach taken in multiple papers using role-played situations (Rashkin et al., 2019), assigned traits (Shuster et al., 2018, assigned movie preferences (Kang et al., 2019), or even an entire fantasy universe (Urbanek et al., 2019;Prabhumoye et al., 2020). Open problems. Relying on role-playing and publicly available data creates a potential distribution mismatch problem, where it is unclear whether people are talking about the same things and in the same way as if they were truly having a normal private one-on-one conversation. This makes the creation of public benchmarks difficult. If improvement of an agent trained on role-played and public data correlates well with increased satisfaction with private conversations, then this would be a good sign that we can keep focusing training efforts on that regime. Another avenue would be to explore using privacy-preserving libraries such as CrypTen 3 and decentralized approaches such as federated learning (Konečný et al., 2016) to handle learning from non-public data. Locally personalizing a shared model (for example, on a personal mobile device) with data that would remained siloed on the personal device could be another way to deploy fine-tuned personalized models in a privacy-preserving way. These solutions would require drastically down-sizing current models and making them efficient and small enough that they could be loaded on device and locally updated without communicating with external servers. Benchmarks could then rely on gauging people's satisfaction on their private interaction with the agent. Our research on more efficient models ) is a step in that direction, and so are works that explore smaller footprints for Transformer-based models, e.g. through knowledge distillation (Sanh et al., 2019), adaptive spans (Sukhbaatar et al., 2019), or layer pruning (Fan, Grave, and Joulin, 2019). Measuring Success Evaluation of Natural Language Generation remains a broadly unsolved problem, with a patchwork of solutions being used across different domains. The open-ended nature of generating sequences in a multi-turn setup naturally makes the task difficult to evaluate -with full evaluation possessing many of the difficulties of the task itself as it requires deep understanding of the content of the conversation. In this section, we describe some of the approaches that have been used to evaluate dialogue systems, their relative advantages, and a number of open problems. Human Evaluations Goal-oriented dialogue systems often have clear evaluation methodologies, e.g. task completion can be measured if the correct actions are taken (Hastie, 2012;Henderson, Thomson, and Williams, 2014;Bordes, Boureau, and Weston, 2017;El Asri et al., 2017;Wen et al., 2017). Chitchat tasks, such as those discussed in this work, are more open ended, and instead feature conversations without a precise goal that can be automatically evaluated. Furthermore, automatic metrics (discussed below), have not been shown to have a clear correlation with human evaluations (Liu et al., 2016;Lowe et al., 2017). This means the current standard for all dialogue research involves human trials. However, there are multiple ways one may choose to evaluate the effectiveness of the system, and human judgements are often difficult to measure. Today, the two most common evaluation forms for dialogue include single-turn pairwise evaluation, and multi-turn Likert evaluation. In single-turn pairwise evaluation (Vinyals and Le, 2015;Li et al., 2016), a human evaluator is typically presented with a full conversational context, and shown two possible responses, and asked to pick which model they feel is better. This test affords the benefits and simplicity of an A/B test, but fails to take into account any multi-turn aspects of a conversation. For example, a model which repeats itself across multiple turns will not be identified by such a system, a behavior known to be highly disliked by human evaluators (See et al., 2019). It furthermore removes any noise produced across multiple turns, wherein a system would be required to ingest its own responses in the conversation history, rather than some produced by a human (Li, Weston, and Roller, 2019). Another common evaluation framework is multi-turn Likert evaluation (Ashwin et al., 2017;Venkatesh et al., 2017;Zhang et al., 2018;Rashkin et al., 2019;See et al., 2019;Dinan et al., 2020Dinan et al., , 2019b, in which a human evaluator is asked to discuss with an agent for several minutes, and then evaluate performance on a Likert (1-5) scale. Such evaluations easily capture a model's ability to carry on longer conversations, and handling of out-of-distribution situations, and therefore may be preferred over single-turn pairwise evaluations. However, multi-turn Likert is not without its own difficulties: it is considerably more labor intensive than A/B tests, as it requires longer and higher-cognitive involvement from the annotators, and it relies on absolute identification rather than relative discrimination, even though absolute identification is not reliable in humans (Stewart, Brown, and Chater, 2005). Likert evaluations are often not strong enough to find statistically significant differences between some models, making it difficult to measure incremental improvements (Kulikov et al., 2019). To make matters worse, it is usually necessary that one must also re-evaluate the baselines at the same time as one's novel model, as the distribution of human annotators can easily shift over time, causing measurement errors (See et al., 2019). Another common difficulty is related to sequential effects (e.g., reviewed in Stewart, Brown, and Chater (2005)), where the first system an annotator evaluates can heavily influence their future ratings, causing difficulties in using an absolute scale. Some groups have proposed hybrid approaches between single-turn pairwise evaluation and multi-turn Likert scoring. For example, Novikova, Dušek, and Rieser (2018) propose a method that combines continuous scales and relative assessments, but in single-turn, rather than multi-turn evaluation; and Adiwardana et al. (2020) propose binary good/bad annotations of individual utterances in a multi-turn setting. Li, Weston, and Roller (2019) recently proposed ACUTE-Eval, in which evaluators are asked to complete pairwise evaluations of complete dialogues. An example of ACUTE is shown in Figure 9. This setup affords a number of advantages over both single-turn pairwise, and multi-turn Likert evaluations. The explicit use of comparisons remedies many of the issues of sequential effects, while still providing the ability to expose issues that are present only in multi-turn evaluations. Furthermore, the pairwise setup facilitates replication and efficient reuse of data: conversations collected in previous trials and by other systems can be directly compared with a new system, without having to recollect additional data. This can significantly reduce the resources needed by a new evaluation, and ensure that multiple papers are comparing to prior work consistently. As a trade-off, ACUTE-Eval does require that one performs two stages of evaluation: one where humans conduct conversation with a model, and another where third-persons indicate pairwise preferences. If one has many systems to compare, this may actually increase resource requirements, since one must pay the full price of multi-turn collection, and another of pairwise evaluations. Fortunately, we can reuse the same dialogue in multiple pairwise comparisons, reducing the number of conversations required to detect statistical significance, alleviating some of the issue. When comparing to multiple existing systems, the benefit of being able to re-use old collections outweighs the resource requirements of the new collections, mitigating these effects (Li, Weston, and Roller, 2019). However, as an alternative, we find that ACUTE-Eval can also work in "self-chat" mode, where models are used for both sides of a conversation, instead of human-model chat. This eliminates the requirement of the initial human collection, and conversations may be generated without human involvement, dramatically reducing the resource requirements of evaluation. We found in our experiments that results from self-chat experiments highly correlated with those of humanchat experiments, for most, but not all systems (Li, Weston, and Roller, 2019). This mirrors other successes in using selfplay, self-chat, and simulated users to evaluate dialogue systems (Fazel-Zarandi et al., 2017;Shah et al., 2018a,b;Wei et al., 2018;Ghandeharioun et al., 2019). Automatic metrics Evaluation of chitchat tasks with automatic metrics is difficult precisely because of their openended nature. For example, the answer to the question "What are you doing tonight?" has many possible answers, each with little word overlap. This means standard metrics based on word-overlap with reference responses, as frequently used in question-answering (Rajpurkar et al., 2016) or machine translation (Papineni et al., 2002), do not work well, and have poor correlation with human judgments (Liu et al., 2016;Novikova et al., 2017;Lowe et al., 2017). Nevertheless, a number of studies do report automatic metrics, sometimes without human studies (Lowe et al., 2015;Serban et al., 2016;Parthasarathi and Pineau, 2018). Some commonly used wordoverlap metrics include F1 (Rajpurkar et al., 2016), BLEU (Papineni et al., 2002;Li et al., 2017c), ROUGE (Lin, 2004;, CIDEr (Vedantam, Zitnick, and Parikh, 2015;Zhou et al., 2020), and METEOR (Banerjee and Lavie, 2005;Zhou et al., 2020). Each covers slightly different aspects, and may be more appropriate in specific situations, but none is known to be a perfect evaluation of conversational models. More specialized metrics may be used for specific subtypes of conversational AI systems. For example, ranking models are often evaluated using Recall @ K or Top-1 Accuracy (Zhang et al., 2018;Dinan et al., 2019bDinan et al., , 2020Humeau et al., 2019). These can be used as rough proxies for improvements, but the open nature of dialogue means that there may be many valid answers in a given candidate list. Such metrics are also unable to capture how well a model will generalize to new situations. Similarly, generative models typically report perplexity of a held-out test set (e.g. Li et al. (2017c); Dinan et al. (2020);Shuster et al. (2020);; Adiwardana et al. (2020); Roller et al. (2020)), and recent work has even found perplexity correlates strongly with human evaluations within the same model class (Adiwardana et al., 2020). While perplexity does give a good estimate of the probability that a generative model would produce the gold label, such results may be actually quite rare under beam search (Fan, Grangier, and Auli, 2018;Holtzman et al., 2019;Welleck et al., 2020), and not representative of an actual generation of a model under beam search or sampling. Perplexity also depends on the dictionary, and not all models will necessarily have entirely comparable perplexities, especially when unknown words are present in validation or test labels, making it difficult to compare systems across multiple time horizons . Modern model using BPE dictionaries further complicate complications of comparing perplexities across multiple systems (Sennrich, Haddow, and Birch, 2016). Specialized systems, which focus on improving specific behaviors of generative models, might instead focus on specialized metrics that are not indicative of overall generation quality, but instead on specialized behavior like repetition (See et al., 2019;Welleck et al., 2020;Li et al., 2020) or vocabulary usage (See et al., 2019;Holtzman et al., 2019;Li et al., 2020). Altering the behavior of the generation method can dramatically influence human evaluations, while maintaining identical or near-identical perplexity (See et al., 2019;Welleck et al., 2020Welleck et al., , 2019Adiwardana et al., 2020;Roller et al., 2020). Noting the inadequacy of each of these automatic metrics, a number of researchers have proposed learning metrics for dialogue evaluation (Lowe et al., 2017;Ghandeharioun et al., 2019). Typically, this is done via a regression from a number of automatically-extracted features (e.g. sentiment and semantic similarity) to human evaluations. In particular, Ghandeharioun et al. (2019) perform such correlations using features extracted via self-play of models. Such systems provide a promise of improved speed of research and development of dialogue agents, but so far have not been met with wide adoption. A common point of criticism is that there can be little effective difference between a learned metric and one that is used as an actual model for performing utterance selection. Put another way, one can easily maximize a metric by employing methods like ranking all utterances according to a learned metric, or using Monte Carlo Tree Search (Kumagai et al., 2016) during generation to naïvely optimize the automatic metric. In this manner, the problem of learning an automatic metric is difficult to disentangle from the rest of dialogue research. Open problems. Selection of an automatic metric for dialogue research, or natural language generation in general, remains a widely open problem that attracts many researchers. Despite concerns, we remain optimistic about methods which approach the problem via learning. Future work may additionally consider holistic evaluations, which require the full conversation to complete before being able to make an individual prediction. This may help mitigate concerns around using the metric to drive dialogue selection. Similarly, adversarial networks may provide a potential avenue for improving automatic selection via continually improved detection of compute-generated responses. In the short term, shared tasks may offer the best avenue to finding automatic metrics which correlate with human judgements (Ashwin et al., 2017;Dinan et al., 2020;Yoshino et al., 2018), but also rely on a diversity of submitted systems in order to consider such evaluations. If all participants use similar models with similar pretraining using similar corpora, then we should not expect clear distinctions to be made apparent in shared tasks. Behavioral Metrics Yet more alternatives are available as models are deployed to real users, especially behavioral metrics. For example, in the Alexa Prize, models were evaluated by how many turns were completed before a conversation was abandoned (Ashwin et al., 2017). Others might be evaluated by the retention rate of users (e.g. how many users choose to have a second or third conversation with a model). Such behavioral metrics can be powerful implicit indicators of preferences of users, but have a large number of issues. For example, models which frequently ask for clarification will naturally result in more turns of conversation, but naturally frustrate users; systems which initiate a conversation will have higher retention, but may not be appreciated by users. A careful and thoughtful balance of allowed behaviors must be employed, and researchers should feel discouraged from using "engagement hacks." Open problems. There is significant question as to what are the correct implicit behavioral metrics to collect, and what few methods exist now depend heavily on the medium and design of the system. As more models are deployed to the wild, we encourage researchers to share their successes and best practices so that the community may come to a consensus. Discussion It is likely that all of the above (human evaluations, automatic metrics, and behavioral metris) and more, will need to be measured with some granularity, in order to understand trade-offs of different behaviors and attributes. In the short term, deployed models should likely focus on retention, in order to ensure a steady stream of users to afford experimentation and iteration. Discussion In this section, we strive to enumerate our core research and ethical values. We discuss how we prioritize trade-offs in our decisions, as well as lessons internalized from our experiences with different steps in the development process. We end on reflections of trends in the community, and calls for action within the community. Values and Guiding Principles One primary principle behind our work is openness. We strive, whenever possible, that the findings of our research should be shared freely via publications whenever it provides benefit. Furthermore, the items necessary to reproduction of our results should additionally be made public when possible. This includes pretrained models, but also code and data necessary to reproduce these results. We believe that siloed research inhibits progress of the field, and point to the recent radical improvements in the NLP community stemming from the openness of the publication of Transformers (Vaswani et al., 2017) and explosion of following open models (Devlin et al., 2019;Lample and Conneau, 2019;Dai et al., 2019;Yang et al., 2019;Liu et al., 2019). With the trend of pretraining coming to dominate the field, open datasets are more important than ever. Our own data, models, and code will are made public via ParlAI 4 , our unified platform for dialogue research. Our current best approach, BlenderBot is available there. Indeed, our unified platform additionally propels our second value: swiftness. In the past few years, the NLP community has been radically changed via massive improvements to the availability of compute and data. Reacting and improving upon the most state-of-the-art work will be important to the success of open-domain conversational agents. We have found ParlAI to be important to remaining swift during these times of rapid development. By providing a unified platform for collection of data, implementation of models, evaluation of agents, and deployment of agents, we are able to significantly reduce development time. For example, our recent development of Polyencoders required no modification to be evaluated using our new evaluation framework (Li, Weston, and Roller, 2019). We also prioritize privacy as a value in our research. Online chats are often where our most intimate and sensitive discussion happens, and one may imagine that users may be even more uninhibited in their interactions with bot. As such, we must act responsibly with respect to data releases. This means that all users must be provided informed consent around how their conversations will be used. Furthermore, we should only release anonymized versions of data, and make every effort to ensure that sensitive data is not included in public releases. Indeed, we must always value privacy over openness and swiftness, whenever our values are in direct conflict with one another. However, we believe that we can have all three values at once: for example, games with role-playing aspects, like Beat-the-Bot, have mitigated the likelihood of sensitive information being included in a conversation, and enable us to open-source the data. In future deployments, we will also add a private mode to appropriate selections, which disables logging and data collection. We also hope that federated learning (Konečný et al., 2016) and other privacy-oriented machine learning techniques will enable future variants to perform learning in a privacy-first manner. Our Experiences We have internalized a number of lessons from experiences with training and releasing models. Pretraining First, we have found that pretraining is important to performance for nearly every variant of chitchat we have experimented with (Wolf et al., 2019b;Dinan et al., 2020), including both in retrieval and generative methods Zhang et al., 2019;Adiwardana et al., 2020;Roller et al., 2020). Furthermore, we have consistently found that domain-specific pretraining is important to high performance: that is, using dialoguelike pretraining significantly outperforms generic pretraining on resources like Wikipedia and BooksCorpus (Dinan et al., 2019b;Humeau et al., 2019;Dinan et al., 2019aDinan et al., , 2020. This further underscores the importance that our models should be openly available, in order to ensure researchers with fewer computational resources are still able to conduct high-quality research. Such efforts are important to ensuring pretraining acts as a rising tide, rather than protectionism for the groups with the most resources. Efficiency Even groups with large computational resources will find that models must be computationally accessible in order to be deployed on a wide scale. Deployed models need to run on commodity hardware, without access to GPUs or TPUs. Indeed, this was the core motivation behind the development of Polyencoders . As a rule of thumb, a researcher should be able to communicate with her model in real-time on a mid-tier laptop, with zero additional development effort. This restriction ensures that we are developing models that are able to be deployed easily. Furthermore, since automatic metrics are untrustworthy in dialogue, it also ensures that a researcher can manually test her model, understanding its power and limitations. Although the recent trend in NLP is to train larger models requiring GPUs for inference Devlin Adiwardana et al. (2020), methods for more efficient inference and smarter algorithms provide ways to retain performance while keeping high performance (Sanh et al., 2019;Fan, Grave, and Joulin, 2019;Humeau et al., 2019). Best practices We have also adopted a number of software engineering best practices around our development process. In particular, we have found automatic testing and continuous integration to be invaluable to our developments. We regularly verify our released models perform as expected, allowing us to readily identify and remedy backwardscompatibility issues. High quality code reviews, even during early model prototypes, have helped us identify bugs and misunderstandings in our models early and sped up development. Universal usage of a shared platform ParlAI has ensured that we maintain a consistent level of quality across multiple projects, and that we can minimize efforts around reproduction and guarantee performance and longevity of models. Many of these benefits are obvious and well-known to software engineers, but are easily forgotten and ignored in research. Deployment In contrast to development of models, deployment has presented us with a very different set of lessons. There are a number of major engineering complications involved whenever games require the involvement of two or more humans, as in Beat-the-Bot; these run antithetical to usual scaling recommendations like sharding. Slowness in pairing can significantly frustrate users, and cause them to abandon the game, further exacerbating the issue. Furthermore, users want the game to react instantly, but wish to take their time in responding. The result is that users may be more satisfied with games and interactions that do not require another human to be present. We have also found that deploying models has resulted in a consistent and steady stream of adversarial users ("trolls"). This is highly unsuprising, as shown by the legacy of Microsoft's Taybot (Neff and Nagy, 2016;Miller, Wolf, and Grodzinsky, 2017). Interestingly, we found that adversarial users had a tendency to assume that they were training the bot online, similar to Microsoft Tay, and believed the bot would learn to mimic their suggested responses quickly. These users focused heavily on suggesting highly offensive responses, especially to innocent and common questions. Other adversaries focused on asking sensitive questions in hopes the bot would produce an offensive response, with the intention of publicizing the bot's failures. Both of these experiences emphasize the importance of the safety of responses, especially around the need for safety-in-context. They also demonstrate the significant risks in online learning, and why it must be deployed with extreme caution and safeguards. Shared Tasks We urge the community to rally behind a definitive set of tasks and measurements for conducting engaging chitchat research. The current state of the fractured community makes it difficult to compare works, despite having similar domain and purpose. Recent competitions, such as the ConvAI2 challenge and the DSTC7 challenge (Yoshino et al., 2018), stand as excellent models for such endeavors. As we progress, these challenges must incorporate more and more difficult challenges encompassing all of the behaviors that are necessary for the ultimate bot. We note that our recently developed DodecaDialogue suite offers an evaluation framework that encompasses many of the research challenges discussed in this document, and encourage the rest of the community to employ it. Such standardized tasks should also include real, interactive learning systems, such as those developed in the Alexa Prize competition (Ashwin et al., 2017). We hope that future iterations will also provide for more liberal open data and participation. We believe these are an excellent way for research to progress, and encourage these continue and expand. Naturally, this is complicated by the number of open research problems, such as what is the correct way to do automatic evaluation of systems. Software As we standardize on data, groups may see additional benefit from standardizing on software stacks as well. Our ParlAI 5 framework attempts to integrate standardized data, released models and software as a one-stop solution and its wide adoption by the community has created an ecosystem of improved support, feature development, and engineering improvements. Our openness to collaborations, contributions, and requests from groups outside our own organization is a reflection of our core belief that sharing research tools is the most productive way to make fast advances as a field. A number of design principles went into ParlAI, which we believe have paid repeated dividends and enabled faster research for ourselves. In particular, ParlAI has focused on unifying the format of all tasks into simple input-output text pairs have helped us to treat ensure our systems are highly reusable, and that different tasks may be easily combined to produce models which exhibit joint behaviors. We also attempt to avoid overly-specialized architectures as often as possible, ensuring our models are also useful across a wide variety of tasks. Furthermore, we observe that today's best models are tomorrow's baselines, and that the centralized repository of reproducible and pretrained models and implementations significantly lowers the overhead needed to perform comparisons to prior work. Perhaps more easily overlooked is the success we have experienced by enforcing all parts of the system act via a unified API of Agents. This means that datasets (teachers), models, crowdworkers, and actual users are all exposed through a unified means. As a result, we may move effortlessly and raplidly progress between Wizard of Oz data collection, model training, human evaluation, and model deployment. Conclusion Our research has shown that it is possible to train models to improve on some of the most common weaknesses of chatbots today. Over time, we'll work toward bringing these subtasks together into one unified intelligent agent by narrowing and eventually closing the gap with human performance. In the future, intelligent chatbots will be capable of open-domain dialogue in a way that's personable, consistent, empathetic, and engaging. As part of our contribution to the broader research community, we're sharing our new models, training code, and data sets within ParlAI , our open source dialogue research platform. We hope that this platform will continue to foster research advances across the research community and contribute to pushing dialogue research forward, addressing many of the open problems we have described here.	2020-06-23T01:00:47.820Z	2020-06-22T00:00:00.000	{ "year": 2020, "sha1": "18c54279a916293153db45e6db8422eaa52539cd", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "18c54279a916293153db45e6db8422eaa52539cd", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Computer Science" ] }
1907176	pes2o/s2orc	v3-fos-license	Gravitation as deduced from submicroscopic quantum mechanics Based on the model of a"soft"cellular space and deterministic quantum mechanics developed previously, the scattering of a free moving particle by structural units of the space -- superparticles -- is studied herein. The process of energy and inert mass transmission from the moving particle to superparticles and hence the creation of elementary excitations of the space -- inertons -- are analyzed in detail. The space crystallite made up around the particle in the degenerate space is shown to play the key role in those processes. A comprehensive analysis of the nature of the origin of gravitation, the particle's gravitational potential 1/r, and the gravitational interaction between material objects is performed. It seems reasonably to say that the main idea of the work may briefly be stated in the words: No motion, no gravity. Introduction Gravitation still remains the most obscure theme in physics. Notwithstanding this, the conceptual foundations of general relativity and quantum mechanics allow concrete modifications in each other in the interface region, Ahluwalia [1]. Describing conceivable new generation quantum-gravity experiments in certain atomic systems, Ahluwalia [2] then notes that they are based on the possibility that quantum gravity might affect the nature of general symmetry, or that the theory of general relativity itself may not provide a complete description of gravitation. Many fundamental problems of gravity and the quantum behavior of the matter have been raised also in remarkable review by Sorkin [3]. In review [4], Dowker and Sorkin investigate some quantum properties of spatial topological geons, particles in 3+1 quantum gravity. Geons allow the construction of spin statistics both fermionic and bosonic and this seems to be a way to an understanding the inner construction of Nature. It is interesting that the authors note that geons are found in some common environment. But what kind of environment? May geons interact with the environment? In particular, what is the law of the motion of the geon? Of course, these and other questions need separate studies. However, the fact of the presence of an environment has engaged a special attention. This signifies that the spacetime, or a vacuum is endowed with structural properties that turn us out to a certain aether, but rather quantum one. In recent years several serious attempts to re-introduce an aether in physics have been made [5][6][7][8][9][10][11]. Similarly, in the previous works of the author [12][13][14][15][16][17][18][19] a simulation of a vacuum in the form of a cellular elastic space has been proposed and deterministic quantum mechanics based on the strong interaction of a moving particle with such a space net has been constructed. The real space has been suggested to consist of peculiar super densely packed superparticles, or balls, which are found in the degenerate state over all the multiplets. A particle is created from a superparticle and hence the particle is interpreted as a local deformation of the degenerate space. Such idea agrees very well with the mathematical study of space carried out by Bounias and Bonaly [20,21] (see also Ref. [22]) who have shown basing on topology and set theory that the necessity of the existence of the empty set leads to the topological spaces resulting in a "physical universe". This allows the investigation of links between physical existence, observability and information. Thus the empty hyperset provides for a formal structure that correlates with the degenerate cell of space and supports conditions for the existence of a universe. Magnon [23] following Bounias also pointed out about some "primordial cell" and "existential principle". Moreover, the surprising thing is that the information on the existence of the supreme substance and the first cause of matter in the form of an indivisible thing is contained in Bhagavad-gītā (see also Bhaktivedanta Swami Prabhupada [24]): "In spite of that material body is subjected to destruction, [the subtle particle] is eternal" [Bg. 2.18]; "It never takes birth and never dies at any time nor does it come into being again when the material body is created. It is birthless, eternal, imperishable and timeless and is inviolable when the body is destroyed" [Bg. 2.20]; "After some time it is disenthralled by entirely annihilation of the material body. Yet it endures the destruction of the material world" [Bg. 2.22]; "It is not fissionable, not burning out, not soluble, and not drying up" [Bg. 2.23]; "Since it is not visible, its entity does not change, its properties remain unchangeable" [Bg. 2.24]; "Yet there is another nature, which is eternal and is transcendental to this manifested and unmanifested matter. It is supreme and is never annihilated. When all in this world is annihilated, that part remains as it is" [Bg. 8.20]. There are also other approaches introducing an aether and those ones that aims to the examination of properties of space-time at very short distances. In particular, Amelino-Camelia notes that the nature of space-time has to show a "fuzziness" at distances close to the Planck one [25] and at these distances particles may be described as geometry "defects" [26]. "Foamy" quantum gravity fluctuations are studied in Refs. [27,28] and such fluctuations seems can be able to modify particle propagation in an observable way [29]. Semi-classical space-time is emerged in canonical quantum gravity in the loop presentation as a polymer-like structure at microscales, which allows a possible correction to the Maxwell equations caused by quantum gravity [30]. Matone [31] has studied the quantum Hamilton-Jacobi equation for a system of two particles and concluded that gravitation in such system had a pure quantum-mechanical origin. Coming back to the author's concept of the constitution of space and the creation of elementary particles in it [12][13][14][15][16][17][18][19], we should note that the approach supposes the formation of a deformation coat around the created particle [12][13][14][15][16]. The coat differs from the degenerate space in that that its superparticles possess mass. Therefore the coat may be called the space crystallite. Its size corresponds to the Compton wavelengthλ 0 of the created particle,λ 0 = h/M 0 c [14]. According to the definition [12,13], the induction of mass means that the volume of a superparticle changes from its volume in the degenerate space. In other words, if we set that a superparticle constricts with deformation, the mass will be defined as the ratio of superparticle's initial and final volumes, m 0 ∝ V/V sup 0 for the massive superparticle and M 0 ∝ V/V par 0 for the particle, where V is the typical volume of a degenerate superparticle and the volumes of a deformed superparticle and a particle are respectively V sup 0 and V par 0 . When a particle moves, the crystallite travels together with it. However superparticles themselves are motionless: the crystallite state migrates by a relay mechanism. The rearrangement of superparticles due to the particle's motion takes place with a velocity that equals or exceeds the speed of light c, but the motion of the particle itself occurs with the velocity v 0 < c (hereinafter v 0 designates the initial particle's velocity, which the particle acquires at a momentary push). The moving particle emits and absorbs elementary excitations of the space -inertons, which appear as a result of friction that the particle undergoes when moves against superclosely packed superparticles. Why do inertons not leave the particle totally? Why do they come backwards again? This is because they carry not only mass but electromagnetic polarization as well as particles are charged. However, this is the subject of a separate study. Here we only point out that the velocity c free of an absolutely free inerton that might migrate in the degenerate space should exceed the velocity of light c, perhaps several times. Accordingly, the initial velocitŷ c of electromagnetic polarized inertons at which they are emitted from the moving particle and accompany it, is not able to reach the threshold value c free and probably c <ĉ < c free . At the same time the mean value of the velocity of the particle's inertons over a period still remains less then the velocity of light c even when the velocity v 0 of the particle approaches to c (in fact 1 2 (ĉ + 0) < c even atĉ = √ c 2 + c 2 , see expression (7) below). Detailed theoretical consideration of the motion of a canonical particle has shown [12][13][14] that owing to the interaction with superparticles the particle looses its kinetic energy on the section λ/2 of the particle path where λ is the amplitude of spatial oscillations of the particle (the de Broglie wavelength). The lost energy is spent on the creation of inertons. On the next section λ/2 the particle absorbing inertons acquires the velocity v 0 , and so on. Thus inertons form a substructure of the matter waves. It should be emphasized that the major theoretical prediction, the existence of clouds of inertons surrounding particles such as electrons and atoms, indeed, has recently been substantiated in a number of experiments [17][18][19]. In addition to friction, or inertia of the space, which results into the generation of inertons, the moving particle undergoes the dynamic pressure on the side of the whole space [13] (the pressure acts on the particle through its coat). The space pressure causes an additional deformation in the particle volume, V part Here we write the speed of light c, because the electromagnetic polarization that accompanies any particle imposes a limitation on the speed of inertons. Since the value of v 0 varies in the spatial interval λ of the particle path, and this is one of the peculiarities of the model constructed, the particle energy should periodically change as well. The energy passing from the particle to its inerton cloud is given by the kinetic energy of the particle 1 2 Mv 2 . At this moment the particle mass changes from Mc 2 to Mc 2 − 1 2 Mv 2 and hence the behavior of the mass obeys the law That is, the value of mass oscillates along the particle path within the spatial period, or amplitude λ. De Broglie [32] was the first to indicate that the corpuscle dynamics was the basis for the wave mechanics. With the variational principle, he obtained and studied the equations of motion of a massive point reasoning from the typical Lagrangian in which, however, the velocity v of the point was constant along a path. The study showed that the dynamics had the characteristics of the dynamics of the particles with a variable proper mass. Oudet [33] conjectured similar peculiarities for the electron. Papini and Wood [34] studied a geometrical solution to the de Broglie variable mass problem. What is more, the de Broglie view is well substantiated as it immediately follows from the Schrödinger equation: the equation giving quantum solutions contains a pure classical parameter -the unchangeable particle mass. Based on the theory developed in recent author's works [12][13][14][15][16][17][18][19], the present paper shows that the creation of inertons from a moving particle is stipulated by the existence of the space crystallite around the particle. The de Broglie's idea and the author's previous results are taken as the starting point. The study includes an extensive description of the processes of the energy and mass transmission from the particle to superparticles when the particle and coming superparticles collide. Besides for the first time the theory arises the question, how is the gravitational potential induced by a particle/object in the ambient space? It is read that such induction is caused by inertons enclosing any material object. Thereby it is the dynamic inerton field that is responsible for the generation of the static Newton potential 1/r. The appearance of the gravitational interaction between both particles and material objects is elucidated in some detail. Emission of inertons For the solution of equations of motion of the particle and inertons the relation has been used in the preceding papers of the author [12][13][14]. In (1) M is the particle mass,Ẋ i is the vector of the particle velocity at the moment of the ith inerton emission (X i is the radius vector of the particle and the dot over X i means the differentiation in respect to the proper time of the particle), m i is the mass of the ith inerton andĉ is its initial velocity. Relation (1) is the consequence of the intersection of geodesics of the particle and the ith inerton. At the moment of the ith collision of the particle and the superparticle, the former emits the inerton, as follows from (1), whose energy is equal to the double kinetic energy of the particle itself. It turns out that a moving object emits an enormous quantity of inertons N within a half-period of its spatial oscillation (N = λ/V 1/3 where λ is the spatial period identical to the de Broglie wavelength and V 1/3 ∼ 10 −30 m, or V 1/3 ∼ l Planck ≈ 10 −35 m, is the suggested size of the superparticle). So the energy of the emitted inerton cloud is of the order of the huge magnitude NMv 2 0 /2. This situation is possible when the ith inerton gains the energy because the medium around the particle is in the vibrating state rather than because of loss in kinetic energy of the particle itself (its energy suffices for the creation of one inerton only [12]). The availability of the crystallite in the space around the particle has been proven theoretically in paper [14]. The crystallite's superparticles are characterized by mass. Hence collective vibrations of massive superparticles, similarly to vibrations of atoms in an ordinary solid crystal, should be inherent to the crystallite. Therefore, vibrating superparticles are able to strike the particle in such a way that the particle will loss energy in collisions and then will generate excitations in the surrounding. Excitations caught by the vibrating superparticles will carry away from the particle. At the formal consideration of quantum gravity, for instance in the model of D-brane string solutions, researchers also face the problem of scattering. Kabat and Pouliot [35] treated the zero-brane dynamics in which one could probe distances much shorter than the string space. Ellis et al. [36] have studied the gravitational recoil effects induced by energetic particles. In their approach defects in spacetime are derived as an energy-dependent refractive index and D-brane foam has corresponded to a minimum-uncertainty wave-packet. The shift of D-brane particle has been induced by the scattering after interaction with a closed-string state. Let us treat now the process of the particle scattering by superparticles in the model discussed in detail. Two relationships were previously derived [13,14]: Here expression (2) connects the spatial period λ of the particle oscillations (the de Broglie wavelength) with the specific enveloping amplitude Λ of the inerton ensem- Figure 1: Rearrangement of the mass state of superparticles around the moving particle. ble. Expression (3) connects the effective size of the crystalliteλ v 0 , defined by the Compton wavelength of the particle, with the specific amplitude Λ of inertons. However the crystallite is dynamic: along the particle's velocity vector permanently occurs the relay readjustment of superparticles from the massless to massive state and again to the massless one (Fig. 1). The resultant of the movement of superparticles is directed antiparallel to the vector v 0 . Thus along the line of the particle motion the state of superparticles changes dynamically: all the time in the coarse of the particle motion superparticles change their state from the degenerate one to the massive one. But in transversal directions the state of superparticles remains practically unaltered: superparticles surrounding the particle continuously save the same massive state, i.e., in these directions superparticles in the crystallite might be considered as hard. Since in any crystal atoms vibrate, massive superparticles should vibrate in the said crystallite as well. However from the pattern above it turns out that superparticles in the crystallite suffer rather pure transversal vibrations -their equilibrium positions transversely vibrate in reference to the vector v 0 . Such transversal vibrations of the crystallite we may call the transversal vibrating mode. The elasticity constant γ of the crystallite and the average mass m cr of a crystallite's superparticle determine the cyclic frequency of these collective vibrations [14] Note that the value of γ is given by the particle mass at rest M 0 , so γ is not universal. Now one can reproduce in detail the picture of the particle motion with the emission and absorption of inertons. Let us assume that at some initial moment, the particle has the velocity v 0 and mass M = M 0 / 1 − v 2 0 /c 2 , equivalently, the particle is characterized by the additional deformation to its own volume V part 0 due to the received velocity v 0 : The particle in motion runs into superparticles of the crystallite, which vibrate in directions transversal to the vector v 0 . The superparticles' vibratory longwave modes can be considered as a longwave excitation of the crystallite: where k v 0 = π/λ v 0 is the wave number and a is the size of a superparticle in the crystallite. The longwave approximation (5) is true not only for v 0 ≪ c but also for the velocity v 0 close to c. The violation of inequality (5) happens only at the up-relativistic velocity v 0 of the particle whenλ v 0 is brought near to a (but this value v 0 is certainly unattainable experimentally). Note that here we have the absolute analogy with the solid because it is known from the theory of solid state (see, e.g. Ref. [37]) that longwave excitations are elastic waves of the medium. The velocity of these waves -the sound velocity v sound -is determined from the relation v sound =ā γ/M , whereā is the lattice constant,M is the atom mass andγ is the elastic constant of a crystal. In this approximation, the group and phase velocities coincide. The vibration energy saved in the crystallite of a moving particle can be found from the equation [14] ω where M = M 0 / 1 − v 2 0 /c 2 is the total mass (we ignore the energy that stemming from the particle spin, see Ref. [14] for details). Essentially, collisions of the particle with the crystallite mode represent an impetuous attack (with the speed of c) to which the particle is subjected on the part of the front of this mode. Each of the ith action of collisions results in the emission of the ith inerton: the mode knocks out the inerton from the particle. The ith inerton has a mass m i and two velocity components:ẋ ⊥ i directed perpendicular to the vector v 0 andẋ i along this vector. It is obvious that the componentẋ i passes over from the particle to the ith inerton; the componentẋ ⊥ i is caused by the momentum of the crystallite mode. Supposing the speed c is the velocity of the crystallite mode, we get relation Let m iĉ 2 be the total energy of the ith inerton. After a series of N collisions of the particle with the mode, the energy of the latter should gradually decrease: Therefore, the mode spends the energy ω i for splitting of the particle and the creation of the inerton that appears with the same energy ω i = m iĉ 2 . Immediately after the ith collision the whole space net readjusts the crystallite to a new quasi-equilibrium state in which the vibration energy ω i+1 corresponds to a new total particle energy M i+1ĉ 2 . And then again a collision occurs and the energy passes from the crystallite mode to the (i + 1)th inerton, etc. until the particle stops after the Nth collision, i.e., when its velocity and inert mass are exhausted, However, how is the inerton mass created? Obviously the process of mass creation takes place at the sacrifice of the particle mass. It should be assumed that at the ith collision of the particle with the mode, the particle loses a portion δV i of its initial relativistic deformation V part 0 1 − v 2 0 /c 2 , or, in other words, the relativistic mass M decreases on a value of the inerton mass m i created on the ith coming superparticle. Thus the particle mass also acquires the index i and the generalization for expression (1) should be the relationship Now let us dwell on the inerton motion. The inerton is created in the elastic medium, i.e. crystallite, which is specified by the elasticity constant γ. The constant γ determines the interaction of the created inerton with the elastic force of the crystallite. As a result of such an interaction, the elastic force aspires to make the inerton return to the particle. Thereby the force sets the inerton into oscillation along the line which is superimposed with the vector of the initial inerton velocity ĉ. The cycle frequency of oscillation of the ith inerton is The maximum distance to which the inerton migrates from the particle is the amplitude Λ i , or Λ i =ĉ/(ω i /2π). Behind the crystallite boundary (in the case when v 0 < c, i.e., when Λ i prevails the Compton wavelengthλ v 0 ) the inerton is guided by the degenerate space whose elasticity is adjusted to the mass of the moving inerton in conformity with relation (9), i.e., γ space = γ = ω 2 i m i . Note once again that γ is not a universal parameter of the universe. This is a consequence of the axiom [12,13] of adiabatic motion of particles/quasi-particles in space when a moving object does not leave faults into the passed range of the cellular space. In this event spatial oscillations of the particle/quasi-particle are exemplified by the adiabatic invariant of space, Planck's constant , and in the case of our inerton this adiabatic invariant can be written as = ( 1 2 m iẋ 2 i )/ω i . On the other hand, in papers [12][13][14] we characterized the ith inerton by the frequency of its collisions with the particle, 1/T i . Therefore we can relate this frequency to the oscillation frequency (9), i.e. ω i = 2π/2T i . The parameter 1/T i provided for the energy exchange between the particle and the ith inerton. It is this parameter that allowed us to obtain the periodicity in solutions for dynamic variables of the particle and inertons. According to these solutions, within even time half-period of collisions T i /2 and the spatial one λ i /2, the particle is scattered by the crystallite vibrations with the subsequent formation of inertons. Within odd half-period the inertons are absorbed by the particle; they transmit the mass and the longitudinal component of the velocity to the particle, i.e., the inertons guide the particle. The guided particle is followed by the restoration of the vibrating crystallite mode, or in other words, the whole degenerate space restores the crystallite state to the initial dynamic state at which the particle possesses the initial velocity vector v 0 . We have stressed that the energy of the emitted inerton is in the direct proportion to the particle energy at the moment of collision with the crystallite mode. Since both the energy of the particle and that of the mode decrease from collision to collision, the same relation should be true for the energy of inertons. It follows that the inequality m i+1 < m i holds for the mass of emitted inertons. The value of mass of inertons, which carry out inert and gravitational properties of particles, has been evaluated in paper [38]. It has been shown that masses of inertons emitted and then absorbed by a moving particle are not strongly fixed but distributed in a wide spectral range much as the photon frequency varies from zero to the frequency of high-level γ-photon. Mass dynamics In our model the canonical particle is considered as a stationary deformed elementary cell, or superparticle, of the real space. Hence the origin of the initial particle velocity v 0 is not a moving point in the space as in contemporary geometry [39] but a cell whose volume is different from that in the degenerate state. Thereby we can decompose real space in two subspaces: the external and internal subspaces. Namely, in the first space the entire cell is considered; it is characterized by an observable trajectory l and the vector of the particle velocity v 0 which sticks out of the particled cell belongs to this subspace as well. The second space represents the cell itself: the cell's size, its inner substructure, degree of its global surface curvature (or deformation), and hidden motion of the cell's centre-of-mass. For example, the notion of spin-1/2 determined as the proper pulsation of the particle [14] should be related to the internal space. Thus a moving particled cell is scattered by surrounding superparticled cells and produces changes in both the external and internal characteristics. So, due to the interaction between a moving particle and oncoming superparticles the particle undergoes a peculiar splitting, or fission. The particle mass decays and is spent for the creation of inertons. Apparently this is the process that falls under the study in the internal space. The particle moves in the external space and its behavior here is characterized by its proper time t = l/v 0 , where l and v 0 is the particle trajectory and the particle velocity respectively. Let us refer all processes which occur in the internal space to the proper time t of the particle. This means that the dynamics of a global cellular deformation of the particled cell, i.e., the dynamics of the inert mass M, should be treated as a function of t. Thus two masses may be distinguished in our system: the particle mass M and the mass of particle's inerton cloud m. Besides we have to operate with the rate of change of mass per time t. For this purpose, let us introduce valuesṀ andṁ, which are the rates of change of the particle inert mass and the inerton cloud mass respectively. Longitudinal mass To describe a dynamics of the particle mass along the particle path l, we need a model Lagrangian written in the form equivalent to the classical one. Let us start from the following specific Lagrangian of the internal space, i.e., the original mass function Here, the first and the second terms are peculiar kinds of 'kinetic energies' of the particle inert mass and its inerton cloud mass respectively; the third term describes the mass exchange; π/T is the cyclic frequency of collisions between the particle and the inerton cloud. The Lagrangian (10) of the internal space is similar in its form to the Lagrangians of the external space used for the particle motion in papers [12,13,16]. The Euler-Lagrange equations of motion are as follows If we pay attention to the fact that at the initial moment t = 0 the total mass was concentrated in the particle, we come to the solution where the amplitude, i.e. the value of the exchange mass is As follows from solutions (13) to (15), the mass of the particle changes periodically along its path from the initial value M = M 0 / 1 − v 2 0 /c 2 in node points, which corresponds to moments of time t = nT where n = 0, 1, 2, ..., to the rest mass M 0 in antinode points. Transversal mass To the careful observer, the solutions (13) -(15) correct only along the particle path l. However, the particle mass obeys changes also in transversal directions and expression (7) indicates conclusively that its change is different from that prescribed by the solution (13) and (15). Indeed, if in the longitudinal direction the velocity of inertons isẋ = v 0 , in transversal directions the velocity is other,ẋ ⊥ = c (7), and, therefore, in the latter case the Lagrangian must also be distinguished from (10). Namely, it has to look as follows where the first and the second terms describe the 'kinetic energies' of the particle proper mass and its inerton cloud mass respectively; the third term describes the mass exchange; π/T is the same cyclic frequency of collisions between the particle and the inerton cloud. The solutions to the Euler-Lagrange equations derived on the basis of the Lagrangian (16) are where the amplitude, i.e. the value of the exchange mass is Consequently, the particle mass is pumped over from the particle to the inerton cloud, i.e. to the ambient space, and then comes back from the cloud to the particle. Thus, the periodical transfer of the particle mass occurs by a tensorial law: it varies from M ≡ M 0 1 − v 2 0 /c 2 to M 0 along the particle pass l and changes from M 0 to 0 in transversal directions. Mass dynamics of inertons In the two previous sections inertons have been treated as quasi-particles which migrate hoping from superparticle to superparticle by a relay mechanism. At the same time, it is the vibrating motion of crystallite's superparticles that knocks inertons out of the particle. Therefore it is reasonable to attempt to consider the splitting of the particle mass and its transformation to the ensemble of inertons from the point of view of wave process. The ith inerton created by the particle starts to migrate having a mass m i . The appearance of the ith inerton means that the superparticle next to the particle takes on the deformation, on the mass m i . In other words, the volume of the corresponding superparticle located at this place is compressed. So the inerton, i.e. the mass m i , starts to barrel in with the initial velocityĉ from superparticle to superparticle. Space gradually brakes the moving inerton and that is why the value of inerton mass m i decreases with r passing to zero at distance Λ i from the particle. At the same time the substrate elastically wrinkles. In other words, the local deformation is sensibly transformed into the rugosity [40] of the cellular space net, which reaches the maximum value at Λ i . Then the space net begins to straighten the rugosity and initiates the reverse inerton motion: the rugosity grades again into the local deformation, the inerton takes on the original mass m i and after all it arrives to the particle and returns the mass m i to it. It stands to reason that for the discrete elastic medium the most straightforward pattern for the motion of the ith inerton from the particle's environment can be presented in Fig. 2. In such a way in the system studied we can distinguish two variables: the inerton mass m i that describes a compressed cell of the space net (Fig 2,a) and the space net rugosity u i that may be regarded as some kind of a cell displacement (Fig. 2,b). The mass m i is viewed as a function of the proper time t i of the ith inerton. The rugosity is supposed is a function of the radius vector x i of the ith inerton and its proper time t i , i.e. u i = u i ( x i , t). Further we will deal with dimensionless variables ̺ i = m i ( x i , t)/m i0 (here m i0 is the initial maximum value of mass of the ith inerton) and ξ i = u i ( x i , t)/(V sup 0 ) 1/3 , which may be considered as two potentials of the inerton. Let us study a dynamics of the system in the framework of the following specific Lagrangian Here, the sum spreads to all emitted inertons; the first two terms are the rates of change of the mass ̺ i and the rugosity ξ i respectively, the third term describes their interaction. In the Lagrangian (16) the parameterĉ specifies the speed of interplay between the two potentials, namely, the mass potential and the rugosity one. The Lagrangian (20) includes the function ∇ ξ i . In this case, the Euler-Lagrange equations take the form (see, e.g. ter Haar [41]): where the functional derivative For the Lagrangian (20) Eqs. (21) and (22) yield the following two equations: These equations being uncoupled change to the following (in general case the right hand sides of Eqs. (25) and (26) are equal to constants C 1 and C 2 , however, the appropriate initial and boundary conditions for ̺ i and ξ i make it possible to put C 1 , C 2 = 0). Eq. (25) is a typical wave equation; in our case it describes the behavior of the mass potential ̺ i of the ith inerton. Thus this characteristic, the mass of the inerton, indeed changes periodically as we initially suggested. Let us treat the solution to equation (25), because it is this equation that is most informative at the study of the mass behavior of the particle's inertons. The initial conditions are The Cauchy problem to Eq. (25) with conditions (27) reduces to the problem on oscillations of a string with the fixed end x i = 0. In our case the solution -in the form of a standing wave -will show how the value of inerton mass decreases along the inerton path, from x i = 0 to x i = Λ i . Now let us turn to the generalized problem, namely, let us treat the behavior of the mass of particle's inerton cloud considered as a single object. The single cloud of inertons is periodically emitted by the particle and then absorbed again. Besides one should go on to the proper time t of the inerton cloud and to the cloud amplitude Λ that is found from relationship (2) (note that such a procedure was already made in papers [13,14] for other problems). We may describe such a system by the equation which is the generalized one to Eq. (25). Here ̺ comprises the total mass potential of the inerton cloud (and hence the total inerton mass (11), i.e. ̺ = m/µ). Apparently, the system studied features the central symmetry in respect to the particle since inertons issue from the particle throughout the azimuth angle 2π around its path (in a truncated view, it may be called the radial symmetry of the cloud around the supposedly motionless particle). Nevertheless the distribution of inerton trajectories, peculiar rays-strings, and the cause of this distribution is not discussed herein. In such a manner we have reduced the problem to the treatment of wave equation (28) that possesses the central symmetry with the initial conditions the second condition here means that the initial mass of the inerton cloud is zero (i.e. no any inert mass around a particle in the beginning). The boundary condition is Because, ̺ = ̺(r, t), then the Laplase operator in the spherical coordinates with the center located in the particle changes to the following Therefore equation (28) takes the form of the equation of radial oscillations whose solutions (with conditions (29) and (30)) are well-known in classical mathematical physics (see, e.g. Refs. [42,43]). In our case the solution has the form ̺(r, t) = C r cos πr 2Λ cos πt 2T (33) and then functions f (r) and F (Λ, t) in the conditions (29) and (30) are equal to The solution (33) can be rewritten explicitly, namely, for two components of the inerton cloud mass, respective longitudinal ( ) and transversal (⊥): where amplitudes are given in expressions (15) and (19), i.e., respectively In equations from (31) to (35) The solution obtained, (33) to (35), directly demonstrates that the particle periodically throws about its mass and then takes it back. The distribution of mass around the particle follows the amplitude of mass oscillation of the inerton cloud µ , ⊥ r cos (πr/2Λ). (36) where (V sup 0 ) 1/3 < r ≤ Λ. But this means that the inerton cloud forms the gravitational potential of the particle! Indeed, the two orthogonal components of mass (36) being distributed around the particle signifies the formation of a mass field around the particle. In other words, the mass field is the result of the defractalization of parts of the particle when its mass gets asymmetrically smeared around the "core" superparticle in the range covered by the inerton cloud amplitude Λ (2). Thus in this range the cloud's superparticles acquire additional deformations, i.e. become massive and their mass decreases with r in compliance with expression (35). Inertons migrating from the particle and then turning backward to it densely fill the environment. Superparticles by which inertons migrate contract and this means that the entire space net around the particle contracts as well. In such a manner any test particle having occurred under the mass field (35) will follow its gradient. That is, it is the contraction of the space net between two massive particles that realizes the attraction between them. This is the inner reason of the phenomenon of gravity, or attraction. In a region confined by the size of a superparticle and the inerton cloud amplitude, (V sup 0 ) 1/3 ≪ r ≪ Λ, and at v 0 ≪ c the time-averaged mass field (35) is reduced to a good approximation Going over to conventional physical units, we have to multiply the both sides of expressions (37), (38) by a factor −G/(V sup 0 ) 1/3 where G is the Newton constant of gravitation; as a results we obtain Thus the longitudinal component (39) of the gravitational potential of a particle depends on its velocity v 0 , though the transversal component (40) exactly represents Newton's gravitational law. Consequently, the introduction of inertons means that a new kind of a mechanics makes its appearance. Actually, in addition to contact and elastic interactions, which are characteristics of classical mechanics, and in addition to the specificity of orthodox quantum mechanics, which repeats the scheme of classical mechanics by using statistical tools, we gain a field mechanics. This one is based on the concept of the elastic tessellation space and, moreover, the concept treats an object as an element of the tessellattice [22]. And just that aspect is responsible for the field mechanics: inertons carry not only the momentum and kinetics energy from one particle to another, but in addition they transfer the local deformation into the surrounding space. When a test particle is thrown into the massive field induced by the other particle, it is contracted and, therefore, the gravitational law (39), (40) prescribes alterations in energetic and force characteristics of the test particle caused by the particle contraction due to its embedding into the contracted tessellation space. Discussion In the quantum substrate studied both the substrate itself and its components -superparticles-cells are elastic. They are those peculiarities that enable one to interpret mass as a deformation of a superparticle. Such a deformation looks like an uniform reduction of the superparticle's volume (though the deformation is fractal [22]). Peculiarities of the particle motion in the space involve the particle motion itself including changes in behavior of center-of-mass, i.e. spin components (↑, ↓), the motion of the particle's deformation coat (equivalently a crystallite) along with the particle, and the creation and migration of inertons. The dynamics of the system under consideration exhibits the periodical decay of the particle mass into mass of inertons. At a distance of Λ from the particle the inert mass carried by the inerton cloud completely turns into the rugosity of the space. Such periodical transformation of the deformation from the particle to the spatial rugosity permits the description of the inerton motion in terms of standing elastic spherical waves. They have been these waves that induce the deformation, i.e. gravitational potential U ∝ M/r (39), (40) surrounding the particle (but the distance r is limited to the enveloping amplitude Λ of the particle's inerton cloud). Therefore, standing spherical inerton waves are carriers of the real gravitational interaction between canonical elementary particles. It is significant that the mutual interaction appears only as a result of the motion of particles because there is no information about an absolutely motionless particle relative to the space beyond the border of the crystallite (see Ref. [14] for details). In such a manner the crystallite is a peculiar kind of the screen between the particle and the degenerate space tessellattice. For instance, the electron's spatial crystallite of the electron (the crystallite shields the electron from the degenerate space) in the state of rest has the size of the Compton wavelengthλ 0 = h/M 0 c ≃ 2.42 × 10 −12 m. The electron of an hydrogen atom is characterized for the lowest level by the following parameters: v 0 ≃ 2.1 × 10 6 m/s, λ = h/M 0 v 0 ≃ 0.35 nm, and the amplitude of the inerton cloud (2) Λ = λc/v 0 ≈ 20 nm. Thus the amplitude of the cloud far exceeds the atom size a H ≃ 0.1 nm, i.e., Λ/a H ∼ 200. The crystallite size of the nucleus of a hydrogen atom is determined by the Compton wavelength of the proton which equals 1.32 × 10 −15 m. This value is much smaller than all the above-mentioned parameters. Hence it is reasonable to conclude that it is not an atomic nucleus that influences the electron, but on the contrary, the field of transversal potential U ⊥ (40) of the electron's spherical inerton wave embraces the nucleus. This is evidently true for the electromagnetic interaction as well, because the electromagnetic field may be regarded as some kind of a polarization, which is superimposed on inertons. What is a radius of the gravitational potential created by a heavy material object placed in the degenerate space? It is apparent that the components of the compound object are in ceaseless motion. For example, in a solid atoms vibrate in the neighborhood of their equilibrium positions. These vibrations result from the atom-atom interaction, which consists of both the elastic component of electromagnetic nature and the inertonic one [18]. Amplitudes of vibrating entities in a solid play a role of the de Broglie wavelengths of these entities [19]. If, say, a solid sphere with a radius R sph consists of N sph atoms and the interatomic distance is a, the spectrum of wavelengths of acoustic waves of the sphere is defined as 2ān where n = 1, 2, ..., N sph /2. The vibrating motion of atoms generates respective inerton clouds in the degenerate space, which move in synchronism with atoms in the acoustic waves. The value of the corresponding amplitude of the nth inerton cloud can be obtained from the relationship for the frequency of collisions of the corresponding acoustic wave with the inerton cloud, i.e. amplitude of inerton cloud Λ n that accompanies the nth acoustic wave is (compare with expression (2)) where v sound is the sound velocity of the sphere. In the classical limit the object size far exceeds its de Broglie wavelength. In that case long wavelength harmonics should determine the inerton field structure around the object and this means that any reasonable laboratory time interval t is still very small in comparison with T , so in expression (35) we may set \| cos(πt/2T )\| ∼ = 1. For instance, if the sphere has the volume 1 cm 3 ,ā = 0.5 nm, N sph = 10 22 atoms, and v sound = 10 3 m/s expression (41) gives for the biggest amplitude: Λ N sph ∼ 10 18 m. Thus the amplitude of the longest inerton wave Λ N sph , which the material sphere generates as a whole is the maximum distance to which the inerton field of the sphere propagates in the form of the standing spherical inerton wave. Thereby, we can write the gravitational potential of a compound spherical object, which is similar to the particle's (40), where r is limited by inequalities R sph ≤ r ≪ Λ N sph . Let us roughly estimate the time necessary for the emitted inertons to return. For simplicity we assumed that the velocity of inertonsĉ = c. Then expressions above yield for the corresponding time, namely, for time wave period: T N sph = Λ N sph /c ∼ 300 years! The longitudinal gravitational component proportional to v 2 0 /c 2 is also available around macroscopic objects, though its manifestation will be treated in a separate work. On the other hand, the description of the quasi-stationary potential (34a) around the object can mathematically be presented in the form of metric tensor components g ij , which are fundamental in Einstein's general relativity, because of the alteration of the size of superparticles surrounding the central object. Such size alteration we associate with the induction of the gradient of the deformation field (i.e. inerton filed). Consequently at the macroscopic approximation the metric tensor of the deformation potential of the degenerate space may really be regarded as an effective gravitational one. We also enlarge on special relativity. The microscopic mechanism forming the basis for Lorentz transformations in the space substrate has been analyzed by the author in paper [13]. In essence, the microscopic theory takes into account that any material object consists of elementary particles and each particle is surrounded by its own deformation coat, equivalently, the crystallite. The behavior of the crystallite has been described in terms of the discrete hydrodynamics where the point is regarded to be equal to the size of the crystallite, i.e., the Compton wavelength λ v 0 . With the whole object, one may introduce some effective crystallite as well. Then the size of the object crystallite will be defined by the object's Compton wavelength λ Comp . However in the limit l obj ≫ λ Comp (for space) where l obj is the object typical size and t ≫ λ Comp /v 0 (for time) where v 0 is the object velocity, the discrete hydrodynamics can be easily replaced by the kinematics of special relativity which correctly depicts all the parameters of the object but does not reflect the actual (microscopic) origin that is covert by the relativity formalism. Concluding remarks Several important aspects of motion of a canonical particle have been studied in the present paper. The study based on the submicroscopic approach has allowed the consideration of behavior of a particle and the surrounding space along the whole particle path. It has been shown that the availability of the deformation coat, i.e. crystallite, whose massive superparticles are found in the ceaseless vibrating motion plays the role of an original generator that knocks inertons out of the particle, as a result of which its mass decays from M 0 / 1 − v 2 0 /c 2 to M 0 along the particle path and from M 0 to zero in transversal directions. Then the elastic space gives back inertons to the particle restoring its total mass. All these changes occur within the spatial period of particle oscillation, i.e., the de Broglie wavelength λ. Thereby we can say such mechanism represents a real perpetual motion machine, which is launched by an initial push transmitting the velocity v 0 to the particle. It is apparent the particle's inerton migrating in the space is specified by coordinates and the velocity and the equation of motion r + γ m r = 0 (43) gives the information about these parameters [12][13][14]. Yet the inerton as a quasiparticle carries the local deformation ̺ whose migration in the degenerate space is subjected to wave equation (25) ̺ tt −ĉ 2 △̺ = 0. A couple of equations (43) and (44) completely describes the behavior of the inerton. The first equation, (43), describes the motion of the "core" of the inerton; this equation enables the submicroscopic interpretation of quantum mechanics. The second one, (44), describes the value of deformation that the inerton transfers from the particle into the tessellattice: the value ̺ decreases by the law of inverse distance 1/r where r is the space between the particle and the inerton (note that r obtained from Eq. (43) is entered into Eq. (44)). Practically speaking, this allows the interpretation of phenomenon of attraction as a contraction of the space tessellattice between material objects. This can readily be visualized by the following pattern. A physical point being embedded into a medium radiates standing acoustic (longitudinal type) radial waves which spread down to the distant spherical surface; hence, the medium vibrates in the range covered by the sphere's radius. A standing wave, as is well-known, is characterized by ranges of stretching and contraction of the medium. Thus our central point should be treated as a node point and the distant spherical surface's points to be antinode points, that is, the medium is contracted in the vicinity of the central point and stretched at the distant surface (see Fig. 2,b). As a result, the gradient of tension will exactly be directed to the central point simulating the phenomenon of classical "static" gravity. Now let r be a space between two particles and let it obeys the inequality r < (Λ 1 + Λ 2 ) where Λ 1(2) are envelope amplitudes of inerton clouds of the two particles. In this situation the particles will fall under the behavior predicted by Newton/Coulomb law. This has been shown in the previous Section. However, such kind of the interaction between canonical particles is correct only when velocities of interacting particles are widely different. If velocities of particles have the same order, the pattern of particle-particle interaction radically changes. In this case the elasticity γ of each of the inerton clouds is approximately identical in value and hence the particles will contact each other through the attractive inerton potential that obeys the harmonic law, 1 2 γ r 2 . Thus if elasticity constants γ 1 and γ 2 of the two inerton clouds have the same order, one can regard that the interaction between the particles is harmonic. (Indeed, let in a many particle system one particle moves towards the other one and let absolute value of their velocities be the same. Denote the mean mass and the mean velocity of particles' inertons as m and v respectively. Then at the distance between the particles lesser (Λ 1 + Λ 2 ) inertons from one cloud begin to contact inertons from the other one. This means that inertons coming from the opposite directions should elastically collide as the absolute value of their momenta is the same, m v . Therefore the two particles owing to their inerton clouds will contact each other much like two elastic balls. The particles repulsing move in opposite directions and again interact with similar particles which return these two to their initial positions). The existence of the harmonic inerton potential in an ensemble of particles has experimentally been substantiated for atoms in the crystal lattice of a metal [18] and quite recently for the KIO 3 ·HIO 3 crystal, in which just the harmonic inerton potential 1 2 γ r 2 of hydrogen atoms provides for their clustering [19]. Besides it has theoretically been shown in paper [44] that it is the inerton potential 1 2 γ r 2 of nucleons that holds them in a nucleus; in other words, the inerton field of nucleons is a real confining field that ensures the nuclei stability. Similarly, one can set the criterion regarding the interaction of classical objects. Let there be two the same solid spheres, which are characterized by a temperature Θ. This temperature is directly defined by the mean kinetic energy of the spheres' atoms. From the relationship 1 2 Mv 2 Θ = 3 2 k B Θ one gets the thermal velocity of vibrating atoms v Θ = 3k B Θ/M. Then one can gain the corresponding de Broglie mean thermal wavelength λ Θ = h/Mv Θ = h/ √ 3k B ΘM and further, in agreement with relationship (2), obtain the mean thermal amplitude of the inerton cloud Λ Θ = hc/(3k B Θ) of the atom. If the distance r between surfaces of the spheres meets the inequality r < 2Λ Θ , a noticeable elastic inerton correlation will be induced between the spheres, i.e. the spheres will suffer the inerton attraction with the potential 1 2 γ r 2 . For example, Λ Θ ∼ 10 µm at the room temperature and Λ Θ ∼ 1 mm at 4 K. As was mentioned above, such inerton correlations in a solid were studied in paper [18]. It is important to note that in the experimental study of short-range gravitational effects one should leave the induced electromagnetic influence out of the gravitational action. In other words, the van der Waals interaction between macroscopic objects (see, e.g. reviews [45][46][47]) should be reduced. (The effect of Casimir force at a distance about several µm has recently been studied theoretically by Lambrecht et al. [48,49]. Then Long et al. [50] based on the method developed in Refs. [48,49] have calculated the Casimir background and concluded that a gravitational-strength Yukawa force should be distinguishable from the Casimir one at the scale of about 3 µm; however, any new gravitational force has not been revealed experimentally in the range between 75 µm and 1 mm.) At r > 2Λ Θ , thermal inerton correlations between objects are absent and this is why one can simulate the interaction based on the notion of the whole inerton cloud that oscillates in the space around each of the objects. In this case, as it has been shown above, the Newton law is realized. A deviation from the Newton law uncovered by general relativity in a macroscopic range needs a separate analysis of the paired interaction of two attracting objects. Nevertheless, the problem does not occur challengeable from the point of view of the developing submicroscopic concept. The manifestation of the inerton field is observed not only in gravitation induced by this field. Inerton waves are directly responsible for the inexplicable force that since the old times has been called the force of inertia. Many times each of us has undergone the influence of this force, which continue to move everybody when he/she tries to stop or turn abruptly. In such cases our own clouds of inertons continue pushing us slightly when we stop after a fast movement. The inerton field solves also the problem of centrifugal force, which so far still remained an understandable phenomenon of classical mechanics! It is obvious from the stated above that the centrifugal force, which acts on a body moving along a curve line, should appear as a response of the space on a centripetal acceleration applied to the body.	2014-10-01T00:00:00.000Z	2002-05-20T00:00:00.000	{ "year": 2002, "sha1": "c0168f902e9194e1d002d71c5d528bb082fde023", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "f904e35bcd5695b2ac7936251bba4881810cd7fa", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
18557328	pes2o/s2orc	v3-fos-license	The psychology of the naturalist. Probably primitive man recognised the value of birds as food well before his mind became capable of understanding his jealousy of their powers of flight. That humans have long considered flying to be desirable is shown by early artists portraying angels, as spiritual super-humans, with functional wings; also, Peoples such as the Assyrians carved bulls complete with wings, thus representing both strength and Mobility. As well as hunting birds and mammals for f?od, early man must have looked for birds' nests. Clutches of eggs, in season, would have been prized as Palatable food and in some tribal areas mystic significance may have been attributed to the markings and colours of the shells. Obviously, birds have always fascinated man. The Old Testament has many references to birds and so has early literature in general, 'n more recent times, Shakespeare's characters mentioned birds as omens, and included them in meta-Phoric and descriptive passages. This is not surprising as Shakespeare wrote in a community which lived close to nature, when even the largest towns, small by Modern standards, were surrounded by woods, farms and open heaths. Now that natural habitats, together with bird and Mammal poDulations, have shrunk alarmingly it is to he expected that the increasing numbers of humans should demand some association with wild life. Man's complex brain has evolved for so long in close contact With nature that abrupt separation can lead to psycho-'?9ical disturbance. Hence, the modern interest in the study of birds and the multiplication of natural history societies could have been anticipated. To achieve a balance the energy of the human psyche, 0r in a conscious state, the ego, requires some interaction with the life of the moor, the forest and the ?stuary. Perhaps in the majority of cases such contact need only be superficial, but deeper involvement is required in other personality types. Today's ego needs the egg; the problem is to conserve the egg and to allow outlets for the necessary psychic drives. The ego 'tself is too destructive. Fundamentally man is selfish and unless the super-ego, the conscience, can instil Understanding of the necessity of protecting the regaining natural environment, the countryside could disappear in its present form. The development of man's higher cerebral functions has permitted aesthetic appreciation as well as curios-'ty. Thus, it is possible to admire the tones of the blue Pigment of a male chalk-hill blue butterfly Lysandra coridon without … Now that natural habitats, together with bird and Mammal poDulations, have shrunk alarmingly it is to he expected that the increasing numbers of humans should demand some association with wild life. Man's complex brain has evolved for so long in close contact With nature that abrupt separation can lead to psycho-'?9ical disturbance. Hence, the modern interest in the study of birds and the multiplication of natural history societies could have been anticipated. To achieve a balance the energy of the human psyche, 0r in a conscious state, the ego, requires some interaction with the life of the moor, the forest and the ?stuary. Perhaps in the majority of cases such contact need only be superficial, but deeper involvement is required in other personality types. Today's ego needs the egg; the problem is to conserve the egg and to allow outlets for the necessary psychic drives. The ego 'tself is too destructive. Fundamentally man is selfish and unless the super-ego, the conscience, can instil Understanding of the necessity of protecting the regaining natural environment, the countryside could disappear in its present form. The development of man's higher cerebral functions has permitted aesthetic appreciation as well as curios-'ty. Thus, it is possible to admire the tones of the blue Pigment of a male chalk-hill blue butterfly Lysandra coridon without studying the dead specimen in the laboratory. Again, the delicate form and shading of the flower of sheepsbit Jasione montana, can best be er>joyed in its setting on downs or coastal waste, but ^an also has the capability of dissecting the plant and jaking pleasure in enquiring into the chemistry of its "fe. As another example, humans can appreciate the camouflage effect of the olive-green and buff mottlings ?f the egg of a sedge warbler Acrocephalus schoeno-baenus, lying deep in its nest of dried grasses low in marsh vegetation; also, man can ask how the pigments became laid down in the shell of the egg and why some female sedge warblers lay eggs marked by black hair-lines. It is as though such eggs have had curved and irregular lines drawn on them in Indian ink by the use of a mapping pen. It can be understood how diffuse or dappled colours become incorporated in the calcium carbonate matrix during the formation of the egg by pigment cells of the bird's shell gland, yet how can a thin and discreet line be inscribed without being smudged? Similarly, man can admire the beauty of the blues and greens found on the bodies of common dor beetles Geotrupes stercorarius and can investigate whether the colours have biological significance. Then man may wonder if the common tiger beetle Cicindela campestris, with its bright green wing-cases, could catch, prey and breed effectively with wings of a duller hue? The Hunting Instinct Nowadays, one of the most popular branches of natural history is ornithology, attracting a considerable number of competent amateur observers. Should such a bird-watcher, on a coastal marsh maybe, unexpectedly sight a rare wader through his binoculars, he gets a thrill of achievement which is probably very similar to that which a primitive hunter felt as his arrow sunk into the flank of a fleeing deer. Certainly hunting is a strong human instinct and man continues to hunt even though, in most cases, the prey is no longer essential as food; fishing, for example, remains as one of the most widely followed sports in Europe. Today, ornithologists hunt in many different ways. Some continue to trap birds, but instead of the catch being plucked for the pot its wing-formula is determined and it is weighed and a ring placed on the leg. Perhaps parasites are looked for amongst the feathers, as a kind of secondary hunting, then the bird is released: the bird-watcher has gratified his hunting instinct. Other ornithologists get pleasure from adding to their list of bird species seen and will travel great distances, like early hunters, to locate their quarry. Patience is necessary for any hunter: many naturalists are interested in bird behaviour and will wait for hours to observe a particular ceremony or mode of flight. Patience is also required by those who try to capture birds on photographic film, and surely a good bird photograph today has more real value than any dead bird. In the same way, the bird-watcher, who is often a bird-listener as well, uses his portable tape-recorder to capture the songs and calls of the birds he studies.*'Again, there is a sense of hunting fulfilment when a new song is recorded and, happily, this can be done with minimal disturbance of the bird and its surroundings. After a successful hunt, man displays his trophies. Ornithologists of the past collected birds' eggs and skins, often exhibiting them with pride to their friends. Fortunately, such days are almost over although the human wish to acquire is as strong now as at the time of the XlXth Century collectors. Now the naturalist must gloat over the reports of his ornithological investigations and his bird photographs or tape-recordings. One object a man might acquire today is a house and, basically, he defends it rather as a bird defends its territory in spring. If an ornithologist returns to find an intruder in his home he gets enraged, and should he discover that his favourite bird books and recordings have been stolen he will be still more enraged. A male song thrush Turdtis philomelos chases other song thrushes from his breeding territory; similarly, man resents strangers entering his house uninvited. Aggressive Urge Acquisitiveness and hunting are associated closely with man's aggressive instinct. No doubt aggression is well modified and controlled by most naturalists, but at times it is a factor which is harmful to the animals they watch. For instance, a selfish photographer may cut back cover round a nest to get a better picture of the fledglings; again, bird-watchers have been known, sadly, to harass a tired and rare migrant to get a more detailed view of its plumage without any thought for its feeding requirements. Further to this, sometimes a naturalist develops an attitude of ambivalence towards birds; thus, he may shoot teal and mallard in the morning and delight in watching nuthatches searching for bark insects in the afternoon. Throughout history man has been aggressive to the extent of killing and wounding his fellows in the rival camp; it is not surprising, then, that occasionally his aggression against birds amounts to real cruelty. The boy who takes a pin and impales nestlings on a tree is, however, probably behaving no worse than a school bully who beats up a timid new boy. Yet cruelty to birds and mammals need not be an inevitable component of man's aggressive urge; so much can be done to educate the young. Children are inquisitive and highly receptive as they develop; hence they can be taught. But if they are left entirely to their own devices even the more intelligent can become aimless, grubby and sometimes annoyingly destructive. Like birds, man cannot be reared without parental care but man's intelligence enables him to be so much more adaptable: he can invent and use machines and can live with success in a snow-bound city or as a nomad in a tropical desert. With such potential mental agility, most children are thrilled by watching birds but the interest can easily be lost; example by adults and subtle guidance will reinforce their enjoyment of nature and their respect for life in the wild. Probably all normal children are embryo naturalists but it is only by education that this can be realised. It is likely that the association of juveniles or adults in gangs or herds increases aggressiveness. This may lead to another danger for birds because the larger a group of ornithologists becomes, the greater is their competitive spirit and zeal. But birds can hardly appreciate this enthusiasm; they cannot thrive when being chased by aggressive humans. Clearly, the mass watching of birds must be very carefully controlled and organised. The solitary or small group bird-watcher sees more of the birds and gets a deeper pleasure from the experience; certainly less disturbance is created for the birds he wants to protect. Of course, man is a sociable animal, in spite of his territorial and aggressive drives, and it is right that ornithologists should form societies for the exchange of bird information, to plan research projects and for educational purposes. It is of interest, however, that members of such clubs, or herds, elect their own officers and so establish a hierarchy, as an ornithological 'peck-order'. Humans are often classed as having an extrovert or introvert personality; usually it is the extrovert who is selected for office as he makes others aware of his capabilities. Bird-watching used to be regarded as a hobby for introverts, but modern ornithology has so many facets and poses so many questions that it has become a hobby of most personality types, as of all social classes. Probably it has always been so. Reproductive Urge One of the most powerful human instincts is the reproductive urge, and there seems to be no more popular subject for the press or television. Currently, it is the fashion to offer a sexual explanation for many human difficulties. For this reason, it may be refreshing to think of ornithology as having little connection with sex or, at any rate, human sex. Yet, apart from birds' reproductive biology, it is possible that birds have sexual symbolism for some humans, in a psychological sense. Birds have long, probing bills and some sing at night in romantic settings; others enter deep nestcavities; these facts could have sexual linkage in the subconscious mind. It could be significant that birds often appear in the dreams and fantasies of disordered persons although, for that matter, interested normal humans often have frequent bird dreams. On a more realistic level, human pair-bonds have been known to start in the natural history society; there is truth in the proverb: 'birds of a feather flock together'. But whether they are paired or not, bird-watchers gratify some of their innate parental drives when they see fledglings reared to the free-flying stage, and especially those from a nest in their own gardens, and perhaps from a nest-box constructed and sited by their own hands. While sexual and aggressive motives are fundamental to man's life, he has evolved to something higher. Human civilisation is dependent on man's appreciation of beauty and a wish to create objects of beauty. Many people take up bird-watching because of their aesthetic sense while others combine this with the added interest of scientific investigation. As was noted previously, man's brain is so versatile that it is quite possible to enjoy the appeal of a charm of goldfinches Carduelis carduelis taking seed from a thistle patch and to get satisfaction from the statistical and mathematical analysis of the numbers involved. Those who have artistic ability must be excited by capturing a bird's movement and vitality of colour in a painting; it seems obvious that no bird artist could be successful unless he had watched birds in their natural habitat. And it is in birds' habitats that the ornithologist can get such mental stimulation. Considering man's basic instincts it is unnatural that families should live in multi-storey blocks of flats in crowded, noisy and smelly towns. To escape to the country, whether in storm or sunshine, and to watch birds feeding and calling along a moorland stream unaccompanied by the roar of motor traffic, gives the city dweller a feeling of contentment not easily obtained in the world today. Natural history is so fascinating a subject that a Psychological fixation on it is quite understandable. Early man lived and hunted in forests and on open Plains; the brain of modern man cannot lose all conection with the past. Some contact with the countryside is necessary if the nervous tensions and strains of busy urban life are to be avoided. What better means is there of doing this than by becoming a bird-watcher, an entomologist or a botanist?	2018-04-03T01:45:35.502Z	1972-04-01T00:00:00.000	{ "year": 1972, "sha1": "bcda08ab10f6cb4917f65387c1144268c6560f80", "oa_license": "CCBY", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "bcda08ab10f6cb4917f65387c1144268c6560f80", "s2fieldsofstudy": [ "Psychology" ], "extfieldsofstudy": [ "Sociology", "Medicine" ] }
237390974	pes2o/s2orc	v3-fos-license	ORTHODONTIC TREATMENT OF SECONDARY DEFORMATIONS IN ADULT PATIENTS WITH DEFECTS OF DENTITION 104 На умовах ліцензії CC BY 4.0 World. USA: IHCW, 2016 [cited 2019 Apr 24]. URL: http://www.ihcworld.com/_protocols/special_stains /nissl.htm 13. Shkodina A. D., Hrinko R. M., Starchenko I. I. Modern conception as to the functional morphology of the olfactory system and its changes under the influence of some exogenous pollutants. The Medical and Ecological Problems. 2019. Vol. 23, No. 3-4. P. 37-40. DOI: https://doi.org/10.31718/mep.2019.23.3-4.09 14. Šijan Gobeljić M., Milić V., Pejnović N., Damjanov N. Chemosensory dysfunction, Oral disorders and Oral health-related quality of life in patients with primary Sjögren's syndrome: comparative cross-sectional study. BMC Oral Health. 2020. 3 Jul. (Vol. 20, No. 1). P. 187. DOI: https://doi.org/10.1186/s12903-020-01169-5 15. Streptococcus pneumoniae infection regulates expression of neurotrophic factors in the olfactory bulb and cultured olfactory ensheathing cells / S. Ruiz-Mendoza et al. Neuroscience. 2016. 11 Mar. (Vol. 317). P. 149-161. DOI: https://doi.org/10.1016/j.neuroscience.2016.01.016 16. Unsupervised method for normalization of hematoxylin-eosin stain in histological images / T. A. A. Tosta et al. Comput Med Imaging Graph. 2019. Oct. (Vol. 77). P. 101646. DOI: https://doi.org/10.1016/j.compmedimag.2019.101646 17. Yuan J., Li Q., Niu R., Wang J. Fluoride exposure decreased learning ability and the expressions of the insulin receptor in male mouse hippocampus and olfactory bulb. Chemosphere. 2019. Vol. 224. P. 71-76. DOI: https://doi.org/10.1016/j.chemosphere.2019.02.113 Abstract. Orthodontic treatment of secondary deformations in adult patients with defects of dentition. Mirchuk B.M., Maksymov Y.V. In the presence of dentition defects there is a complex of morphological, aesthetic and functional changes that significantly complicate the process of diagnosis and treatment of this pathology. Dentition defects, combined with various dental anomalies and deformations very often impede rational prosthetics and at times make it impossible at all. The aim is to increase the effectiveness of orthodontic treatment of secondary deformations in patients with dentition defects by using primary dental implants as an additional skeletal support. For clinical evaluation of the effectiveness of treatment of secondary deformations in 20 patients with partial dentition defects there was performed orthodontic treatment using a straight arc technique with metal braces of the Roth system and primary detailed one-component implants as an additional skeletal support. On the diagnostic models of the jaws, the mesodistal dimensions of the lost teeth and their possible position in the dentition were determined, the shape of the dentition by methods of Pon and Korkhaus, the position of the teeth in the area of dentition defects was analyzed and the jigs of occlusion according to Andrews were defined. The use of primary dental implants as an additional skeletal support in the area of dentition defects makes it possible to control the rotation of the teeth and at the same time to use orthodontic forces of different intensity during their distal or mesial movement. As a result of orthodontic treatment of secondary deformations, we managed to achieve positive changes in the normalization of angles of dental inclination that limit the defect in patients with dentition defects. Along with the normalization of the angles of dental inclination (torque and angulation) which limit dentition defects we have noticed an increase in the distance between these teeth, which allows to restore dentition defect with dentures, better corresponding to the size of the lost teeth. Important, in our opinion, is the possibility, when using a primary dental implant as an additional skeletal support, to use the technique of segmental braces. The use of primary dental implants in the area of the dentition defect as an additional skeletal support makes it possible to restore angulation and torque of the teeth, which limit the defect, using orthodontic forces of different intensity. As a result of orthodontic treatment of secondary deformations, the distance between teeth limiting dentition defects on the upper jaw increased on average by 2.39 mm (p<0.001) and on the lower jaw -by 2.57 mm (p<0.001). Partial tooth loss is one of the most common forms of dento-mxilllar defects in adulthood. The number of patients with dentition defects in the total population of Ukraine ranges from 70% to 95%. [1,3,4,8]. In young and middle-aged people bounded dentition defects are much more common as compared with the distally borderless defects, and by localization, bounded dentition defects prevail in the posterior triangle of neck in the absence of one or two teeth [7]. At the same time, many authors pay attention to the increase in the prevalence of secondary dentomaxillar deformations, the cause of which is a decrease in the height of the clinical crowns of masticatory teeth due to their destruction by the carious process and dentition defects which make up На умовах ліцензії CC BY 4.0 from 4.8% to 54.5% [11]. As a result of dentition defects presence, a complex of morphological, aesthetic and functional changes arises, which significantly complicate the process of diagnosis and treatment of this pathology. Dentition defects combined with a variety of dento-maxillar anomalies very often impede rational prosthetics, and sometimes make it impossible at all. The aim of the study is to increase the effectiveness of orthodontic treatment of secondary deformations in patients with dentition defects by using primary dental implants as an additional skeletal support. MATERIALS AND METHODS OF RESEARCH For clinical evaluation of the effectiveness of treatment of secondary deformations, 20 patients with partial dentition defects underwent orthodontic treatment by straight-wire technique with the use of metal braces of Roth system and primary detailed one-component VKtemp implants of the native producer VITAPLANT® ("Vitadent Ltd" Ukraine, Zaporizhzhia) as an additional skeletal support. The primary one-component VKtemp implant has a root shape and angulation thread of 15 degrees. The diameter is 2.5 mm and the length is 10 mm, 12 mm, 14 mm or 16 mm. Depending on the type of bone, pilot twistdrills with a diameter of 2.0 mm and 1.5 mm are used, both classical and transgingival implantation is possible. Twistdrills are tapped in the area of the defect using a surgical template. For the fixing of braces on primary dental implants, plastic crowns with supporting orthodontic elements (rings or braces) are made [6]. All patients were diagnosed, treatment plan was made and prosthetic treatment of defects with fixed or non-fixed dentures, orthodontic treatment of secondary deformations was proposed. In each patient, before treatment and after, C-silicone imprints of the maxilla and mandible were made, gypsum diagnostic models were casted. On the diagnostic models of the jaws the mesodistal dimensions of the lost teeth and their possible position in the dentition were determined [12], the analysis of the shape of denture according to Pont [14] and Korkhaus [14], the position of the teeth in the area of dentition defects were analyzed, Andrews occlusion jigs were defined [14]. The results of the research were processed by modern statistical methods of analysis on a personal computer using the standard suite of programs Microsoft Office 2010 (Microsoft Excel [5]) and "STATISTICA® for Windows 6.0" [10] (StatSoft Inc., USA, license 46 # AXXR712D833214FAN5 ) based on the Windows 10 operating system, and by using the NumPy (BSD License), SciPy (BSD License), pandas (BSD License), pandas-profiling (MIT License) libraries, to visualize the processed data, the matplotlib library (BSD License) for Python programming language was used [9]. The hypothesis of the normality of the distribution of the studied indicators was tested using the Shapiro-Wilk criterion and the Kolmogorov-Smirnov consistency criterion. The Kolmogorov -Smirnov homogeneity criterion was used to test the hypothesis that two independent samples belong to one distribution law. Arithmetic mean (M) and standard error (± m) were calculated. The statistical significance of the intergroup differences according to the data obtained was established using the parametric Student's t-test () and the non-parametric U-Whitney-Mann test (p **) [2]. The results obtained were compared between the median and the quartile range of Me (Q1; Q3). We used 3 levels of statistical significance of the differences in the study results -p<0.05; p<0.01; and p<0.001. Studies have been conducted in compliance with ethical standards of scientific medical research involving human beings. RESULTS AND DISCUSSION Dentition defects were diagnosed according to Kennedy classification [13]: on the maxilla -there were 5.0% of patients with class I, 65.0% -with class III and 5.0% -with class IV; on the mandible -10.0% of patients with class I, 16.0% -with class II, 70.0% -with class III. The position of the teeth in the area of defects by Andrews occlusion jigs was determined: angulation (II jig) and torque (III jig) disorder was diagnosed in all (100%) patients with dentition defects, 55.0%without dental turn (IV jig) both on the maxilla and mandible, the absence of interdental space (V jig) on the maxilla was detected in 50.0% of patients and on the mandible -in 30.0%. As a result of orthodontic treatment of secondary deformations in patients with dentition defects, positive changes regarding normalization of angulation of teeth that limit the defect have been achieved. Due to distalization of teeth (teeth that distally limit defect), the mesio-distal angulation decreased on average by 4.15±0.5° (p<0.05) on the maxilla and 12.89±0.5° (p<0.001) on the mandible) as compared to the measurements made before treatment. Mesialization of the teeth that mesially limit the defect caused the increase in the angulation angle on average by 4.66±0.5° (p<0.05) on the maxilla, by 7.56±0.5° (p<0.001) -on the mandible) as compared to the measurements made before treatment (Table 1). A similar positive effect was noted on determining vestibulo-oral angle (torque) of the angulation, which limits dentition defect. Thus, at an excessive value of torque, reduction in the anglulation of teeth was on average at 4.55±0.5° (p<0.01) on the maxilla and at 7.72±0.5° (p<0.001) on the mandible) and at a reduced value of the torque, increase was 4.69±0.5° (p<0.01) on the maxilla and 5.27±0.5° (p<0.001) on the mandible, as compared with values before the onset of orthodontic treatment ( Table 2). T a b l e 2 The average values of reduction and increase of angulation of teeth in the area of dentition defect (M±m) Along with the normalization of angulation (torque and angulation) of the teeth which limit dentition defects, we noticed an increase in the distance between these teeth, this allows to replace dentition defect with the dentures which correspond better to the size of the teeth lost (Table 3). На умовах ліцензії CC BY 4.0 After orthodontic treatment, the distance between the teeth that limit the dentition defect on the maxilla increased on average by 2.39 mm (p<0.001) and on the mandible -by 2.57 mm (p<0.001). The use of primary dental implants as an additional skeletal support in the area of dentition defects allows to control the rotation of the teeth and at the same time to use different values of orthodontic forces during their distal or mesial movement (Fig. 1). Thus, for mesialization of teeth (in most cases of premolars) smaller forces (light or medium opening springs) than for distalisation of molars (medium or strong opening springs) were used. In addition, it is known that tooth mesialization is faster than distalization, so when reaching the desired position of the mesially located tooth we could stabilize its position and continue the distalization of the tooth, where a primary dental implant served as a support. In no case, as with the simultaneous use of springs of different force for mesialization and distalization of the teeth, and for the migration of only one tooth (distally or mesially), we did not observe the change of position of the primary dental implant, which served as an additional skeletal support. In case of impared angulation of the teeth that limit dentition defect, the use of a primary dental implant as an additional skeletal support together with bracket systems ensures the restoration of the correct position of the premolar teeth within 4-6 weeks, molar teeth -within for 8-9 weeks. Distalization of the molar teeth lasts 12.5 weeks on average, and mesialization of premolar teeth -7.5 weeks. It is of importance, in our opinion, is a possibility when using a temporary dental implant as an additional skeletal support, to use the technique of segmental bracket-systems (fixing braces on several teeth on the side of defect) in patients with defict of space in the area of defect by changing the position of the teeth (mainly disorders of torque and angulation), which limit dentition defect while retaining the signs of physiological occlusion (Fig. 2). During the entire period of orthodontic treatment there were no cases of change of position or disintegration of the primary implant. We also did not observe inflammatory processes around implants (peri-implants). After the completion of active orthodontic treatment and restoration of the dentition defect, various methods of retention were used: increasing the size of the plastic crown (by photopolymer directly in the oral cavity, or making a new primary plastic crown) until close contact with adjacent teeth; manufacture of removable retentive apparatus with artificial teeth in the area of the dentition defect. 109 21/ Том XXVI / 2 The use of primary dental implants in the area of the dentition defect as an additional skeletal support allows to restore the angulation and torque of the teeth, which limit the defect, using orthodontic forces of different intensity. 2. As a result of orthodontic treatment of secondary deformations, the distance between the teeth that limit dentition defects on the maxilla increased on average by 2.39 mm (p<0.001) and on the mandible -by 2.57 mm (p<0.001).	2021-09-01T15:15:02.679Z	2021-01-01T00:00:00.000	{ "year": 2021, "sha1": "c9eb5271d15b2bf20e030f2bba66a5538f348e59", "oa_license": "CCBY", "oa_url": "http://journals.uran.ua/index.php/2307-0404/article/download/234633/241136", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "ab42c0e3d03b8c31843ad038768abe6f0da874de", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [] }
245037752	pes2o/s2orc	v3-fos-license	Integration of Proteomics and Metabonomics in Exploring the Protecting Mechanism of Gushukang Capsule in Osteoporosis Rats Background Gushukang (GSK) capsule is a Chinese patent medicine for the treatment of osteoporosis (OP). It has been widely used in clinics. However, the specific mechanism and target of GSK in the treatment of osteoporosis is not clear, which needs further study. Methods Metabolomics (GC/MS) and proteomics (TMT-LC-MC/MC) together with bioinformatics (KEGG pathway enrichment), correlation analysis (pearson correlation matrix) and joint pathway analysis (Metabo Analyst) were employed to discover the underlying mechanisms of GSK. Results The regulations of differential proteins Cant1, Gstz1, Aldh3b1, Bid and Slc1a3 in the common metabolic pathway of differential proteins and metabolites between GSK/OP and OP/SHAM were corrected in GSK group. The regulations of 12 metabolites (Tyramine、Thymidine、Deoxycytidine、Cytosine、L-Aspartate and so on) were differential in the common enrichment metabolic pathway between GSK /OVX and OVX/SHAM. Differential proteins and metabolites jointly regulate 11 metabolic pathways, such as purine metabolism, pyrimidine metabolism, histidine metabolism, beta-Alanine metabolism and so on. Conclusion GSK may protect bone metabolism in osteoporosis rats by affecting nucleotide metabolism, amino acid metabolism and immune system. Introduction Osteoporosis (OP) is a systemic metabolic disease characterized by low bone mass, micro-structural damage of bone tissue, decreased bone strength and increased bone fragility [1]. Its clinical manifestations are mainly pain and spinal curvatures [2]. Traditional Chinese medicine (TCM) has been used in China and other Asian countries for thousands of years [3]. At present, under the guidance of syndrome differentiation and treatment, TCM has obvious curative effect and advantages in the treatment of osteoporosis, which can better improve the side effects and safety of Western medicine. TCM formulas not only reduce bone loss by decreasing bone resorption, but also increase bone formation in the multicomponent and multitarget pattern [4]. It can also regulate the overall function of the human body and relieve back pain and lumbago [5]. GSK capsule is a Chinese patent medicine for the treatment of osteoporosis. Its main effective components include epimedium, Rehmannia glutinosa, astragalus, etc. It has the effects of dredging blood vessels, tonifying kidney and Qi, strengthening bone, and so on. GSK capsule has the characteristics of convenient administration, stable curative effect, and small side effects and has been widely used in the clinic [6]. Clinical research shows that GSK is effective in the treatment of primary osteoporosis, which is characterized by the prevention and treatment of OP by adjusting the whole body function. GSK can inhibit the formation of osteoclasts and stimulate the formation of osteoblasts in ovariectomized mice [7], and signi cantly increase the metabolism of vitamin D and calcium [8]. BMP-Smads signaling pathway is important in bone remodeling. Another study found that GSK signi cantly enhanced BMP-2/Smads signal pathway and improved bone microstructure by upregulating relevant osteogenic factors in osteoporosis rats [9]. H-type blood vessels have shown the ability to induce angiogenesis and bone formation [10]. Recent studies demonstrated that GSK can enhance hypoxia-inducible factor-1 α to induce H-type vascular formation and bone formation [11]. In recent years, combined multiomics technology has been widely used in the study of various disease states [12]. Integrating multi-omics data analysis can make up for the data problems caused by data loss, noise, and other factors in single omics data analysis, integrate various interactions isolated at the gene level or protein level, various metabolic and regulatory pathways, and jointly clarify the overall state of the biological system [13]. Among them, the combined application of proteomics and metabolomics technology undoubtedly has great advantages [14]. The data of metabonomic analysis can provide information on the exercise of molecular function for the results of proteomic research and provide clues to explore the regulatory relationship between them. The resulting protein-metabolite interaction network intuitively shows the maladjusted related pathways in the disease state or the overall changes after drug treatment. Proteomics and metabolomics provide us with large information of differential proteins and metabolites, which provides a new research strategy for the target research of traditional Chinese medicine in the treatment of diseases. However, the speci c mechanism and target of GSK in OVX rats are still unclear. Therefore, based on the integrated analysis of proteomics and metabolomics, this study used GSK capsule to treat osteoporosis rats, to further explore the mechanism of GSK in the treatment of osteoporosis. Materials And Methods Animal model establishment 36 3-month-old SPF Sprague Dawley female rats were obtained from Shanghai Slake Laboratory Animal Co., Ltd., and raised to 6 months old in the Experimental Center of Comparative Medicine, Fujian Academy of Chinese Medical Sciences. All rats were maintained in a 12 h light-dark cycle, with controlled temperature (22-24℃) and humidity (50-60%). After one week of adaptive feeding, 36 6-month-old rats were randomly divided into GSK group (n = 12), model group (n = 12), and sham operation group (n = 12). The osteoporosis model was established by removing the bilateral ovaries of rats; the same volume of adipose tissue beside the ovary was removed in the sham operation group [15]. Intramuscular injection of penicillin 800 thousand units / day was used to anti-infection 3 days after operation. There are 12 weeks to develop osteopenia in OVX. GSK treatment We opened up GSK capsule (Z20060270, Liaoning Kangchen Pharmaceutical Co., Ltd.) and mixed it with normal saline by gavage once a day (4.32g / (kg * d)). Sham group and model group were given equal volume of normal saline once a day for 12 weeks. After 12 weeks of treatment, Rats in each group were anesthetized with 2% Pentobarbital Sodium. The rst and second lumbar vertebrae were separated and quickly put into liquid nitrogen. The left tibia was used for micro-CT. Metabolomic data processing The original data was converted into. mzXML format by ProteoWizard, and then peak alignment, retention time correction, and peak area extraction are performed by XCMS program. The metabolite structure identi cation uses accurate mass matching (<25 ppm) and secondary spectrum matching methods and searches the laboratory's self-built database. After Pareto-scaling pretreatment, the data was analyzed by multivariate statistics, including unsupervised principal component analysis (PCA), supervised partial least squares (PLS-DA) and orthogonal partial least squares (OPLS-DA).To further screen the differential metabolites, we used PLS-DA data model to obtain the model Variable Importance for the Projection (VIP) under the condition of positive and negative ion mode to screen the differential metabolites between GSK group / model group and model group / sham operation group. The metabolites with multidimensional statistical analysis VIP > 1 and univariate statistical analysis p value < 0.05 were selected to identify the potential metabolites with signi cant differences. Subsequently, metaboanalyst 5.0 was used to analyze the metabolic pathways involved in differential metabolites and evaluate their importance. When the impact value of metabolic pathways is greater than 0.2, we consider these pathways as potential target metabolic pathways in this experiment. Micro-CT evaluation The proximal end of left tibia was evaluated with micro-CT plain scan and BMD (Fig. 1) to characterize the effects of treatment on bone mineral density (BMD). Compared with the sham group, the model group had signi cantly lower BMD (p<0.001), decreased number of bone trabeculae, increased Tb.Sp (p<0.01), and disrupted bone microarchitecture. Compared with the OVX group, the BMD of GSK was higher (p<0.05), and saved bone mineral loss from OVX in trabecular bone. Results of proteomic analysis The expression of differential proteins in GSK, OVX, and SHAM group were statistically analyzed. Differentially expressed proteins in OVX rats were corrected in GSK group (Top 5)( Table 1). And there were top5 up-regulation and down-regulation differentially proteins in OVX group (Table 2). Results of metabolomics analysis The results of the principal component PCA analysis show that QC and other samples are well aggregated within the group, indicating that the overall quality of the experimental data meets the analysis requirements. There were differences in metabolite levels among GSK, SHAM, and OVX groups (Fig. 2). The signi cant difference metabolites mainly involved in these pathways (Fig. 3) The number of metabolic pathways involved in differential protein and differential metabolites The differential proteins in GSK/OVX group were involved in 70 metabolic pathways (Fig. 4a), the differential metabolites were involved in 73 metabolic pathways, and there are 45 pathways that the two Page 8/21 parts jointly participate in. The differential proteins in OVX/SHAM group were involved in 93 metabolic pathways (Fig. 4b), the differential metabolites were involved in 65 metabolic pathways, and the two parts were jointly involved in 26 pathways. Analysis of enriched differential proteins and differential metabolites in common metabolic pathways According to the statistics of the common metabolic pathways after GSK treatment and castration, differential proteins and metabolites jointly regulate 11 metabolic pathways, such as Purine metabolism, Pyrimidine metabolism, Histidine metabolism, beta-Alanine metabolism, In ammatory mediator regulation of TRP channels, Platelet activation, Tyrosine metabolism, Pathways in cancer, Phenylalanine metabolism, Glutamatergic synapse, Gap junction and so on ( Fig. 5 and Table 3) Table 3 Differential proteins and metabolites enriched by the common metabolic pathways in GSK/OVX and OVX/SHAM. *It is a protein or metabolite with differential expression trends between GSK / OP and OP / SHAM Correlation analysis of signi cant difference proteins and signi cant difference metabolites Based on the Pearson correlation method, the correlation coe cient between signi cant difference proteins and signi cant metabolites was calculated, and the metabolites and proteins with signi cant differences at the key nodes in the network were screened. We constructed a correlation network of proteins and metabolites(\| R \| ≥ 0.5 and p < 0.05) ( Table 4). We found that the metabolites in Histidine metabolism and β-Alanine metabolism were strongly related to the differential proteins. In the correlation network of GSK/OP, seven proteins and eight metabolites as nodes included two correlations and six negative correlations (Fig. 6a). In the correlation network of OP/SHAM, two proteins and two metabolites as nodes included two negative correlations (Fig. 6b) Discussion With the increasing incidence of osteoporosis year by year, osteoporosis has become a worldwide health problem and social public problem [17]. Exploring the mechanism of GSK and other TCM can provide greater advantages and theoretical supplement for TCM treatment and prevention of osteoporosis. Clearly, proteins and small molecule metabolites are closely associated with the occurrence of diseases, thus the combination of proteomics and metabonomics are matched our attention. It has been reported that the speci c mechanism of GSK in the prevention and treatment of OP has not yet been explained, and its molecular mechanism has not been clari ed. Based on the combined analysis of proteomics and metabonomics, the results showed that the expression trend of differential proteins Cant1, Gstz1, Aldh3b1, Bid, and Slc1a3 in the common metabolic pathway of differential proteins and metabolites between GSK/OP and OP/SHAM was different, that is, they were corrected in GSK group. Moreover, Cant1 was involved in purine metabolism and pyrimidine metabolism, and Aldh3b1 was involved in tyrosine metabolism, histidine metabolism, beta-Alanine metabolism and phenylalanine metabolism.After GSK treatment, Slc1a3 was surely correlated with Lhistidine (P=0.045), Aldh3b1 was signi cantly correlated with 1-stearyl-sn-glycerol 3-phosphocholine (P=0.045). Combined with proteomic analysis, Gstz1 and Aldh3b1 were located in the functional nodes of protein-protein interaction network. In addition, the expression trend of 12 metabolites (Tyramine Thymidine Deoxycytidine Cytosine L-Aspartate Ergothioneine L-Histidine L-Anserine P-antothenate Arachidonic Acid (peroxide free) Benzoic acid DL-Phenylalanine) was differential in the common enrichment metabolic pathway between GSK /OVX and OVX/SHAM. The above metabolic pathways are primarily involved in nucleotide metabolism, amino acid metabolism, immune system, cell process, and other signaling pathways, indicating that GSK may treat osteoporosis in rats through these. Uric acid is the nal product of purine metabolism in the human body, which is mainly decomposed by enzymes from nucleic acids and other purine compounds metabolized by cells and purines in food. Uric acid is a reducing substance in the human body, which participates in the redox reactions and scavenging oxygen free radicals, thereby inhibiting oxidative stress [18,19]. A number of studies have reported that serum uric acid (UA) is involved in the pathogenesis of OP by affecting oxidative stress and in ammatory cascades [20]. Higher serum UA levels appear to be protective for bone loss in peri-and postmenopausal women [21]. Cant1, a calcium-activated nucleotide, is essential for glycosaminoglycan synthesis in cartilage [22]. In this study, the key differential protein Cant1 was involved in purine metabolism and was down-regulated after GSK treatment. Amino acid metabolism is an important part of life activities. As the basic unit of macromolecular protein, it is essential for the normal metabolism of the human body [23]. Previous studies [24,25] showed that Amino acid metabolism is important to osteoporosis. In the correlation analysis, the signi cant differential proteins and metabolites have a strong correlation with Histidine metabolism and β-alanine metabolism. Histidine is an essential amino acid in the human body during childhood. In adulthood, the human body can synthesize it by itself and histidine becomes a nonessential amino acid. Histamine is formed under the action of histidine decarboxylase and is a factor that regulates in ammation and allergic reactions [26]. An increase in the amount of bone cortex and bone minerals was found in mice de cient in histamine synthesis, revealing that the lack of histamine may increase the bone density and bone formation of histamine de cient mice by stimulating the synthesis of calcitriol [27]. The results showed that after treatment with GSK, Aldh3b1 in the Histidine metabolism pathway was downregulated, L-Aspartate was downregulated, and Ergothioneine, L-Histidine, and L-Anserine were upregulated. β-Alanine plays a key role in bone metabolism, mainly by improving the production of insulin and insulin-like growth factor-1, as well as the synthesis of collagen and muscle proteins. Studies [28] have shown that β-Alanine may be a potential biomarker of osteoporosis, and β-Alanine levels increase signi cantly in postmenopausal osteoporosis patients with low bone density. Tyrosine is an important amino acid in the body and the main raw material for the synthesis of thyroxine [29]. In previous studies, it has been found that GSK may participate in bone metabolism by regulating tyrosine metabolism and differential proteins. In addition, Aldh3b1 is also involved in the metabolism of Tyrosine metabolism, Histidine metabolism, beta-Alanine metabolism, and Phenylalanine metabolism. We speculate that Aldh3b1 may be closely related to the metabolism of amino acids. Aldh3b1 is a member of the ALDH family. Its function is related to the scavenging of reactive oxygen species [30]. In postmenopausal women, the level of estrogen and antioxidants decreased with age, and the ROS accumulated in the body could not be cleared in time, which induced the occurrence of oxidative stress and led to the damage of osteoblasts and osteocytes [31,32]. ALDH can reduce oxidative stress through a variety of aldehyde metabolism [33]. In this study, the expression trend of arachidonic acid (AA) in the platelet activation pathway and in ammatory mediator regulation of TRP channels changed after GSK treatment. AA and its products play an important role in regulating in ammatory response [34], vascular elasticity, platelet activation, or bone remodeling. AA belongs to polyunsaturated fatty acids (PUFAs), one of the indispensable free fatty acids in the human body. According to the position of the rst double bond position [35], PUFAs can be divided into n-3pufas and n-6pufas, and n-6pufas mainly include linoleic acid and AA. Recent research highlights the potential role of PUFA in the in ammatory regulation of bone remodeling via several cellular pathways [36][37][38]. The combined use of n-3 PUFA and E2 exerted synergistic bone-protective e cacy through upregulation of RUNX2, an essential transcription factor for bone formation, as well as the suppression of bone-resorbing cytokine IL-1β [39]. Conclusion This study reports the signi cant changes of proteins and metabolites in osteoporosis rats treated with GSK capsule. Integrated proteomic and metabonomics analysis showed that the molecular mechanism of GSK regulation might be related to Purine metabolism, Pyrimidine metabolism, Histidine metabolism, betaalanine metabolism, In ammatory mediator regulation of TRP channels, platelet activation, thyrosine metabolism, pathways in cancer. These metabolic pathways are mainly involved in nucleotide metabolism, amino acid metabolism, immune system, cellular processes, and other signaling pathways. This study provides a reference for the molecular mechanism of the GSK capsule in the treatment of osteoporosis. Follow-up studies will also verify the differential proteins and metabolites in these metabolic pathways, to provide a basis for this study. Availability of data and materials The data analyzed during the study are available from the corresponding author on reasonable request. Ethical approval and consent to participate The experimental protocol was approved by the Animal Experiment Ethics Committee of Fujian Academy of Chinese Medicine.The experimental procedure followed the International Association of Veterinary Medical Editors' Consensus of Authors' Guide on Animal Ethics and Welfare and local and national regulations. Consent for publication Not applicable. Map of signi cant metabolic pathways involved in differential metabolites Common metabolic pathways of differential proteins and metabolites in GSK/OVX and OVX/SHAM. Red is the upregulated expression of differential proteins or metabolites, green is downregulated expression of proteins or metabolites, and yellow is a metabolic pathway involved in differential metabolites and differential proteins. Figure 6 Correlation analysis network between GSK/OP and OP/SHAM for signi cantly differential proteins and metabolites	2021-12-12T17:39:53.174Z	2021-12-07T00:00:00.000	{ "year": 2021, "sha1": "aa75eb94022a2bef65d05c16419dd7017193b865", "oa_license": "CCBY", "oa_url": "https://www.researchsquare.com/article/rs-1140478/latest.pdf", "oa_status": "GREEN", "pdf_src": "Adhoc", "pdf_hash": "f9fa0be293c4a472d5ce33188e20ea1c1c56f384", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [] }
237918170	pes2o/s2orc	v3-fos-license	Financial burden of catastrophic health expenditure on households with chronic diseases: financial ratio analysis Background The financial status of households is vulnerable to chronic diseases which entail high medical expenses and income loss. Financial strain can be assessed by four indicators: a household surplus indicator, the liquid asset/debt ratio, a solvency indicator, and a liquidity indicator. We investigated the association between catastrophic health expenditure (CHE) and financial ratio indicators in households with chronic diseases in South Korea. Methods This study applied thresholds to the financial ratios to determine the financial strain. We conducted multiple logistic regression analyses to determine whether CHE is associated with financial strain. Furthermore, we analyzed the relationship between CHE and basic financial indicators, absolute finance size, using multiple linear regression. Results When CHE occurred, all financial ratio indicators deteriorated. However, this was not due to decreases in the absolute size of wealth and income, but rather the relative balance between finances. In particular, the loss of liquid assets was a major factor in the deterioration. In addition, all types of labor-related income deteriorated; only private transfer income increased. Conclusions This study revealed that CHE in households with chronic diseases negatively impacts household finances. It was found that financial coping strategies are only resource consuming. Supplementary Information The online version contains supplementary material available at 10.1186/s12913-022-07922-6. . The economic impacts of chronic diseases include not only direct costs but also indirect costs. To address these direct and indirect costs, households implement "financial coping strategies. " These strategies include the use of savings, borrowing from relatives or acquaintances, liability/loan borrowing, and the disposal of assets, which is defined as probable future economic benefits obtained or controlled by a particular entity as a result of past transactions or events, such as housing, land, and vehicles. Previous studies have argued that such financial strategies serve as buffers against economic crises [7][8][9]. However, simply taking out liability/loans and disposing of assets cannot prevent economic ruin and poverty. First, if one uses liability or borrowing without considering one's ability to repay after a deterioration in health, it will lead to inextinguishable liability in the future [10]. If households fail to fulfill their obligations, they would become delinquent borrowers. Those labeled as such could not freely engage in social and economic activities and face various actions, including crime, that encourages liability fulfillment [11]. Second, Koreans usually regard assets as a means of bearing the costs of children's education, marriage funding, maintaining life after retirement, and investment in asset growth. Therefore, if one disposed of assets due to spending on medical expenses, he would fail to achieve the objectives mentioned above and experience an economic burden. Furthermore, the disposal of assets makes the use of liability/loans more challenging because their availability varies depending on the size of the asset. Many prior studies analyzing the impact of CHE on household finances were limited primarily to examining the effect of income on poverty. In the case of low-and middle-income countries, income could be a proxy for the whole household economy. However, in high-income countries, the sizes of assets and liability markets are also large, and only examining the effect on income is insufficient. We can use the basic financial indicators, which are comprised of each absolute size of assets, incomes, expenses, and liabilities in this situation [12,13]. These indicators can evaluate the association between CHE and household finance more explicitly from multiple aspects. However, finances in households are more complicated. Even if assets are substantial, an individual might go bankrupt when there is not enough cash to make the immediate repayments. To understand these complex effects, scholars in the personal finance field often suggest using financial ratio analysis [12][13][14][15]. Financial ratio analysis is a method that identifies the strengths and weaknesses of households through relative assessment and evaluates whether households are doing well in achieving their financial goals over time by using several financial ratio indicators [16,17]. The Methods section will explain the specific indicators. Therefore, this study examines whether CHE is associated with the increase of financial strain in households with chronic disease, and financial coping strategies would prove only to consume more resources. Furthermore, to understand better how chronic disease impacts indirect costs, we classified income types as labor-related and non-labor-related and analyzed them. To achieve this objective, we conducted several analyses, as follows. • Compared the incidence rate of CHE in households with and without chronic diseases. ▪ Hypothesis 1: The incidence rate of CHE is higher in households with chronic diseases than in those without. • Analyzed the association between CHE and several financial ratio indicators: a surplus indicator, the liquid asset/debt ratio (LADR), a solvency indicator, and a liquidity indicator, which are measured by the ratio of two basic financial indicators (among the absolute size of financial assets, liability, income, and living expenses). See the Methods section for descriptions of the specific indicators and calculation methods. ▪ Hypothesis 2: There are associations between CHE and the deterioration in financial ratio indicators. • Analyzed the association between CHE and several basic financial indicators and each type of income (Labor-related or non-related). ▪ Hypothesis 3: There are associations between CHE and the deterioration in basic financial indicators. ▪ Hypothesis 4: There are associations between CHE and the decrease in labor-related incomes. ▪ Hypothesis 5: There is no association between CHE and non-labor-related incomes. Data This study utilized data from the Korea Welfare Panel Study (KoWePS), which covers all cities and provinces in Korea. The KoWePS evaluates samples using a two-stage stratified cluster sampling design and collects data from household financial records and receipts of OOP to avoid recall bias. The KoWePS provides various data, such as demographic and socioeconomic characteristics of individuals, OOP, assets, liabilities, and income. We mainly used data from 2014, but we added data from 2015 and 2016 as needed. Specifically, since there may have been reverse causality between the independent and dependent variables, we first set the independent variable as the data from 2014 and the dependent variable as the data from 2015. Second, the KoWePS collects data from the relevant year; however, when it comes to the income and living expense variables, it investigates data from the previous year. This is because the KoWePS survey period is so long that the response time of each subject varies. Some may have been investigated in May and others in September. In addition, if the survey is conducted in May, income and living expenses cannot represent for one full year. However, other variables, such as assets or liability, gender, and education level, do not need to consider response time, so it investigated the data of the relevant year. Therefore, we conducted time matching. For example, as mentioned above, we set the independent variable (CHE) as the data from 2014 and the dependent variable (i.e., a surplus indicator) as the data from 2015. When calculating CHE, we used the OOP variable of 2014 and the income variable from 2015. At the same time, when calculating the surplus indicator, we used the living expenses and income variable from 2016. The subjects of this study were households with chronic diseases. The KoWePS collects data on whether they have such diseases or take medicine for more than six months. If at least one person had one or more chronic diseases, we defined the household to which that person belonged as a chronic disease household. Accordingly, 6,270 households participated in the survey for three years, and 4,802 households had chronic diseases. We used publicly available and reliable secondary data from the KoWePS, which were provided through the de-identified samples. The investigator of the KoWePS visited the participants' households, met the respondents, and gained informed consent. Approval of IRB exemption for this study was granted by the Korea University Institutional Review Board (approval number: KUIRB-2019-0214-01), given the retrospective nature of the study. Measures of CHE We calculated the independent variable CHE according to the method of Wagstaff and van Doorslaer (2003) [14]. We divided OOP by household payment ability (disposable income-food expenses) and transformed it into a binary variable [18,19]. When its value was greater than the threshold (i.e., 10%, 20%, 30%, or 40%), CHE was Yes (1), and No (0) otherwise. In the multiple logistic and linear analyses, we used a representative threshold of 10% because it is the approximate level at which households would sell their assets to pay for OOP, pay off liability, and reduce the cost of living other than health expenditure [20]. The threshold level is a commonly used value in empirical studies conducted to date [1]. The OOP of the KoWePS included costs for hospitalization, outpatient care, dental care, surgery (including implants and cosmetic surgery), medicine, nursing care, postpartum care, health checkups, and healthcare supplies (eyeglasses, contact lenses, etc.). Measures of financial strain indicators There are two types of financial strain indicator for households: basic financial indicators and financial ratio indicators [12,13]. First, basic financial indicators are comprised of the absolute size of each financial component: disposable income, earned income, total living expenses with and without OOP, savings, total assets, liquid assets, non-liquid assets, and the total liabilities of households [13,21]. In terms of income, we classified it as labor-related income (earned and business incomes) and non-labor income (property, private transfer, public transfer, and other incomes) to determine whether CHE in households with chronic disease conferred a detrimental effect on individuals' work activities. The basic financial indicators facilitate objective comparison and evaluation; however, they limit assessing the complex aspects of household financial conditions. A financial ratio indicator allows us to better understand these complexities [14]. Financial ratio indicators are measured as the ratio of two financial elements, and a threshold can be set to determine whether there is a financial strain. As financial advisors recommend using several ratios [14], we included the following four indicators: Surplus indicator The surplus indicator is a primary indicator for evaluating the adequacy of household cash flow that can help determine whether the consumption propensity is fine and whether there is the possibility of a deficit in a household [22]. In this study, we measured the surplus indicator as the annual "total living expenses/disposable income" and applied a threshold of 70%, based on Yang et al. (2013) [13]. If the value exceeded 70%, we categorized it as "dangerous. " LADR The LADR is the concept of the ability to liquidate households' total debts with their liquid assets. Liquid assets include deposit and installment savings of bank and stock bonds. Total assets include liquid assets and non-current assets, such as real estate (houses, non-residential buildings, land), agricultural machinery, livestock products (cows, pigs, chickens, etc.), and current car quotations. We defined the bad condition of LADR as liquid assets lower than 20% of their liabilities, as suggested by Griffith (1985) [23]. Solvency indicator The solvency indicator is an indicator of a household's overall financial condition, which determines whether they have total assets to sustain their debts [17,22]. This indicator is represented by "total assets/total liabilities. " Total liabilities include loans from financial institutions, general debentures, card debt, credit, and deposits on leases. We defined this as dangerous if the ratio was over 40% [13,17,22]. Liquidity indicator The liquidity indicator assesses whether liquid assets can sustain the previous standard of living when income is temporarily suspended [13]. It can show whether cash flows and assets are well balanced. It is calculated as "liquid assets/disposable income" and is classified as a bad condition if the ratio is lower than 400%, based on DeVaney (1993) and Yang et al. (2013) [12,17]. Control variables We chose the control variables based on previous studies [24]: household characteristics (number of household members, type of medical insurance, presence of disabled individuals, children younger than 20, elderly individuals aged 65 or older), and the characteristics of the householders (gender, age group, educational level, marital status, employment status). Since the independent and dependent variables in this study were household units, we did not include health-related variables, such as the type of disease and self-reported health, because these are individual units. Statistical analysis This study performed three analyses. First, we compared the CHE incidence rates in households with and without chronic diseases using the chi-square test. At this time, we represented various thresholds (10%, 20%, 30%, and 40%) to examine the sensitivity of CHE incidence. Second, we conducted multiple logistic regression analyses to identify the relationship between CHE and financial ratio indicators in households with chronic diseases. We transformed every dependent variable (financial ratio) to a binary variable, as good (0) or bad (1), according to the thresholds suggested in previous studies. However, since the financial ratio was calculated as a ratio, it was necessary to provide further shape to the model [25]. It is difficult to determine whether the increase in the proportion of the surplus indicator is due to decreased income or increased living costs [14]. Third, we used multiple linear regression analyses to analyze the relationship between CHE and basic financial indicators. All the basic financial indicators often had a non-Gaussian distribution, and because the distribution of the error term is unlikely to be a normal distribution, we applied a logarithmic transformation [14,25]. The variance inflation factor (VIF) was used to test for multicollinearity among the independent variables. Because all the values of the VIF were less than 10, multicollinearity could be ignored. To resolve heteroskedasticity, we computed robust standard errors (detailed explanations are available at https:// www. stata. com/ search/). We used the statistical software program Stata version 14.0 (Stata-Corp, College Station, Texas, US). General characteristics The characteristics of the samples are presented in Table 1. It is notable that most households were unemployed (46.1%), and most of the households had elderly residents (67.4%). These results would have stemmed from the characteristics related to chronic diseases in the KoWePS; as in the total sample of 6,270, the unemployment rate was already 36.1%, and 55.1% of households had an elderly resident. Additionally, in the results for the type of national health insurance, employees totaled 63%, which contrasts with the unemployed rate of the household head. This can be interpreted as many household heads being retired and other household members working instead (Table 1). Incidence rates of CHE The incidence rates of CHE are shown in Table 2. Rates were higher in households with chronic disease. For a typical threshold of 10%, the CHE incidence was only 6.5% in households without chronic disease, while it was 4.4 times higher in households with chronic disease (28.9%). The CHE incidence rates in households with chronic diseases were also high at all thresholds. In addition, thresholds greater than 20% were not suitable for various analyses because of the small sample size ( Table 2). Table 3 shows the association between CHE and financial strain, analyzed using multiple logistic and linear regression analyses. When CHE (threshold 10%) occurred, all financial ratio indicators (surplus, liquidity, solvency indicator, and LADR) deteriorated significantly. In addition, there were differences among the basic financial indicators. First, the CHE had an odds ratio (OR) of 3.248 (p < 0.000) on the surplus indicator, which means that the indicator deteriorated. Further, as a result of further analysis to identify the cause of the deterioration in the indicators, CHE significantly decreased the logarithm of disposable income (coefficient = -0.164; this means that 15.1% of income decreased when converted from logarithm) and increased the total living costs (coefficient = 0.058; this means a 6.0% increase). In this case, the ratio value increased because of the decreasing denominator and increasing numerator. Moreover, we analyzed the logarithm of the cost of living, excluding OOP, as a dependent variable. As a result, it decreased significantly (coefficient = -0.096; this means a 9.2% reduction) when CHE occurred (below the supplementary cell). Association between CHE and financial strain in households with chronic disease Second, CHE correlated with the deterioration of the LADR (OR = 1.301; p < 0.05), which suggests that CHE weakens a household's capacity to hold liability. CHE did not affect total liability, but significantly reduced liquid assets (coefficient = -0.383; 31.8% decrease). Third, CHE was related with a worsening of the solvency index (OR = 1.448; p < 0.01), which may have aggravated household finances in the long term and at the macro level. This indicator is composed of total liabilities and assets. CHE did not affect total liabilities but lowered the logarithm of total assets (coefficient = -0.146; this means a 13.6% decrease). We further classified the types of assets into liquid and non-liquid assets to identify how asset reduction occurs. As a result, CHE had no significant impact on non-liquid assets. Instead, CHE was associated with reduced liquid assets. Fourth, CHE correlated with the exacerbation of the liquidity indicator as much as an OR of 1.375. This indicator is the ratio of disposable income to liquid assets. Since CHE conferred a negative effect on both sides, it was difficult to determine which one had an impact. However, it seems that the decrease in liquid assets is more considerable than income (disposable income coefficient = -0.164; 15% decrease, liquid asset coefficient = -0.383; 31.8% reduction) ( Table 3). Table 4 shows that even in households with chronic diseases, CHE still correlated with the decrease in labor-related income, which is "earned income" (coefficient = -0.385; this means a 32% decrease) and "business income" (coefficient = -0.418; this means a 34.2% decrease). On the other hand, of all income types, the only increase occurred in private transfer income (coefficient = 0.41; this means a 50.7% increase). On the contrary, public transfer income, which is paid by the social security system, decreased inversely (coefficient = -0.143; this means a 13.4% decrease) ( Table 4). Discussion The purpose of this study was to determine whether households with chronic diseases are vulnerable to CHE and to examine whether CHE is associated with the increase of financial strain in households with chronic diseases. Consequently, the incidence rate of CHE was found to be higher in households with chronic disease than in those without as presented in Table 2. This is because households with chronic diseases incur a large OOP, and their income decreases because of health deterioration. In Korea, specifically, since the national health insurance coverage is not sufficient, CHE is highly likely to occur in households with chronic diseases [3]. Furthermore, when CHE occurred in households with chronic disease, almost all financial strain indicators deteriorated in the following year. This study involved two main analyses. First, we conducted logistic regressions to evaluate the financial ratio indicators applied to the threshold model. Second, we analyzed linear regression to determine the components of the ratio indicators affected. We analyzed four financial ratio indicators: surplus, LADR, solvency, and liquidity. To summarize the results, the surplus indicator deteriorated. This could interrupt cash flow. In an additional analysis, it was found that income loss and an increase in living costs affected the surplus indicator. If households with chronic diseases have CHE, the patient may not be able to work, and even other household members may give up work to take care of the patient, which could reduce their income. Furthermore, it seems that medical expenses increased the total living costs. However, the total living expenses without OOP decreased. It can be interpreted that households reduce other consumption to pay for OOP. Second, the LADR deteriorated, which means that CHE weakens a household's capacity to hold liability. The LADR is an assessment of the amount of liability that can be repaid through liquid assets that can be mobilized immediately in the event of an unavoidable situation in which a household must repay all of its liability at once [13,23,26]. LADR is the ratio of total liabilities to liquid assets. As a result of the further analysis, CHE did not affect total liability, but significantly reduced liquid assets. It can be interpreted that households pay for medical costs by withdrawing liquid assets and not borrowing [19,27]. The solvency indicator had an enormous financial scale compared to the other indicators, but the results showed that it associated with CHE. This indicator reflects the full capacity to hold household liabilities, so that it should be alert. Further analysis showed that CHE had no significant relationship with non-liquid assets. Instead, CHE was associated with reduced liquid assets. It can be interpreted that CHE occurs suddenly, and individuals rely on financial assets available for immediate use. Many people view the use of liquid assets as a financial coping strategy, but it is not, as this will eventually lead to a decline in household finances. Finally, the liquidity indicator also deteriorated, which means that households lose the ability to cope with unexpected economic problems. The increase in the liquidity indicator indicates that the money needed to maintain the present economic condition will disappear. In this case, households would become very unstable, both psychologically and economically. A further analysis to identify the cause showed that CHE conferred a negative effect on both sides; however, the decrease in liquid assets was higher. From the results of the four financial ratio analyses, we can deduce several facts, as follows: First, the likelihood of household financial strain is high if CHE occurs in households with chronic diseases. This is not due to a decrease in absolute wealth or income or an increase in debt but rather because CHE broke the relative balance among financial elements. Second, the main factor in the deterioration of the financial ratio index was a decrease in liquid assets. This can be interpreted as people using liquid assets rather than debt or non-liquid assets when CHE occurs. Since three of the four indicators (LADR, solvency, and liquidity indicators) contain liquid assets, this affects all indicators in a chain. Liquid assets and income are different. Income is "newly incoming money, " and liquid assets are "accumulated money" through income. For example, if an employee uses an outpatient service or even pays for expensive medicine, it will not reduce the incoming money (i.e., the salary) but reduce the accumulated money. If OOP were to affect income, then it might be the case that a patient's illness (OOP implied) is so severe that they may not be able to maintain work activities and will lose their wages. Previous studies that analyzed only the impact of CHE on incomebased poverty did not point out any differences between incoming and accumulated money. Third, CHE correlated with the increase in the total cost of living and decreased income, which led to the deterioration of the surplus indicator. Previous studies used the surplus indicator as a tool for short-term evaluation [12,13]; however, the deterioration of the indicator can develop into a long-term problem when it comes to chronic diseases because it implies a loss of workability. To determine whether income reduction occurs through labor, we analyzed the association between CHE and income type (labor-related and non-labor-related). As a result, labor-related income declined. Since CHE represents a more severe status of chronic diseases, it can be interpreted that CHE negatively affects the workability of household members. This interpretation is supported by the fact that non-labor-related income was not affected. Of all income types, the only increase occurred in private transfer income. This can be considered as borrowing money from nearby acquaintances to offset OOP and reduce earned income. In other words, Korean people rely more heavily on social networks than the social security system, which must protect household finances by providing public transfer income. However, private transfer income may not be sufficient to cover the decreasing disposable income. In addition, even if a household reduces expenditure on other goods or services, the total living expenses remain elevated because of OOP. As living expenses increase and income decreases, household cash flows show a deficit. Limitations This study has some limitations. This study could not present relative risks by analyzing modified Poisson regression. Relative risks have an advantage in interpreting more intuitively than the OR. However, it was difficult to use because the prerequisites for the causal relationship of Poisson regression could not be met. This study has a limitation in that it cannot analyze whether CHE has a causal effect on financial indicators, only the correlation investigated. To analyze the causal impact, it is necessary to measure the difference in financial indicators before and after the occurrence of CHE. However, it is difficult to calculate the difference because the financial indicators are measured as a ratio. Moreover, there is a problem that selection bias may occur as the number of samples decreases during the calculation process. In addition, the relationship between the CHE indicator and surplus indicator could have endogeneity. For example, households with a strong consumption propensity will be high in both indicators, and households with a low consumption propensity will be low. Furthermore, this study used the family units of the independent and dependent variables. Therefore, we could not use individual units of the control variables, such as self-reported health. We suggest that subsequent studies utilize panel analysis, considering that chronic diseases affect people over the long term. In particular, liability may have a positive impact on maintaining consumption in households in the short term, but it would likely become an economic burden to repay over the long term. In other words, since the economic burden of liability occurs over time, we must observe a long-term impact. Moreover, income, assets, and liabilities are influenced by the characteristics of households that are not observed; however, a panel analysis could control for such effects. Conclusions Industrialization and globalization have created large economies worldwide alongside aging of the world population as a whole. Accordingly, we are entering the era of chronic disease pandemics [4,[28][29][30]. Chronic diseases are challenging to cure, leading to an economic crisis for households over a long period. In addition, advances in medical technology and pharmaceutical innovation due to industrialization and globalization have accelerated medical expenses [31]. In the past, chronic diseases were regarded as "prosperity" diseases in advanced countries [31]. Now, however, they are permeating middle-and low-income countries, so that medical expenses are expected to skyrocket in the future worldwide [32]. Therefore, the issue of chronic disease is not limited to Korea but affects the whole global population, and it is important to deal wisely with financial strain brought on by diseases. Existing studies argue that exploiting liability, borrowing, and using savings is a financial coping strategy that can protect households from poverty [8,9]; however, this strategy breaks the balance of financial fitness in households. In this study, we demonstrated that several financial strain indicators have deteriorated. These are important facts for an economically advanced country like Korea because one of the reasons for the bankruptcy of households in the Korean community is that they fail to balance the liability/assets ratio. There was no significant increase in liability in this study; however, the relative size of assets could be more critical than the absolute liability size. Even a small liability, if a household is incapable of servicing it, will cause an over liability situation [10]. If assets are reduced, their ability to maintain liability decreases. Households can be classified as credit-impaired in the event of a default and may become bankrupt [33]. A financial ratio analysis was conducted to understand this point.	2021-09-01T15:12:34.419Z	2021-06-22T00:00:00.000	{ "year": 2022, "sha1": "1b2fc5c41706cb8c6906c8804984c71981a082c0", "oa_license": "CCBY", "oa_url": "https://bmchealthservres.biomedcentral.com/track/pdf/10.1186/s12913-022-07922-6", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "dd737a7b2a6c6c1e351e698219d9454ae479610d", "s2fieldsofstudy": [ "Economics", "Medicine" ], "extfieldsofstudy": [ "Medicine", "Business" ] }
17840148	pes2o/s2orc	v3-fos-license	The Three-Dimensional Jump Conditions for the Stokes Equations with Discontinuous Viscosity, Singular Forces, and an Incompressible Interface The three-dimensional jump conditions for the pressure and velocity fields, up to the second normal derivative,across an incompressible/inextensible interface in the Stokes regime are derived herein. The fluid viscosity is only piecewise continuous in the domain while the embedded interface exerts singular forces on the surround fluids. This gives rise to discontinuous solutions in the pressure and velocity field. These jump conditions are required to develop accurate numerical methods, such as the Immersed Interface Method, for the solutions of the Stokes equations in such situations. 1. Introduction.Biological system containing cells can be considered as a multiphase fluid problem, with the cell membranes forming a boundary between fluids of differing material properties.Unlike standard multiphase fluid problems, such as droplets or bubbles, biological multiphase systems, such as red blood cells surrounded by blood plasma, are characterized by inextensible membranes and singular forces due to both a surface-tension like contribution and a resistance to bending [16].Various techniques have been developed to model such systems.These techniques can be split into two categories, based on how they track the location of the interface.Lagrangian methods, such as the boundary integral method [24,25] explicitly track the location of the interface.While these method can be highly accurate, they are difficult to extend to three-dimensions and are not well suited when the membrane undergoes large deformations.Eulerian method, on the other hand, implicitly track the location of the interface.Examples of Eulerian methods include the phase-field method [2,13] and the level set method [18,19].Advantages of the Eulerian methods include easy extensibility to three dimensions and the ability to handle large membrane deformations naturally.The major disadvantage is the added computational cost of implicit membrane tracking. Eulerian methods can be further divided into two categories depending on how they treat the singular forces arising at the embedded membrane.The first type, characterized by the immersed boundary method [15] or the continuum surface force method [4], distribute (smooth) the singular forces over a small region near the interface.This essentially turns the membrane forces into localized body-force terms, which are then included in the fluid governing equations.For most situations this smoothing results in a first-order accurate method for the fluid equations [23]. The second type of Eulerian method avoids the smearing of the interface by explicitly incorporating the singular forces and embedded boundary conditions in the field equations.These forces and immersed boundary conditions are included by explicitly considering the jump in the solution and derivatives of the solution across the interface.Mayo [14] used this idea to solve the Poisson and Biharmonic equation on irregular domains.Leveque and Li called this technique the Immersed Interface Method (IIM) and used it to solve Elliptic equations [9].They later extended this to Stokes flow with elastic boundaries or surface tension [8].Later the immersed interface method was further developed to model the Navier-Stokes equations [7,12,27].Interested readers are referred to the book by Li and Ito for more information about the immersed interface method [11]. The application of the immersed interface method to a multiphase fluid system requires that the jumps in the velocity and pressure field be explicitly calculated.For standard multiphase systems these jump conditions for a discontinuous viscosity across the interface have been determined both in two dimensions [22,23] and three dimensions [5,28].Only recently has work been done on extending the immersed interface method to multiphase flows with inextensible membranes [21].This recent work, though, is limited to the two-dimensional, constant-viscosity case and is not applicable to the more general discontinuous viscosity situation. In this work the three-dimensional jump conditions for a multiphase Stokes flow system with an inextensible interface, singular forces, and a discontinuous viscosity are for the first time derived.The jump up to the second normal derivative of both the pressure and velocity fields will be developed.These jump conditions will be used to construct an Immersed Interface solver.Several analytic test cases are also developed to verify the accuracy of the jump conditions.This paper is organized as follows.First, the basic governing equations for the three-dimensional Stokes system with an inextensible membrane are presented.Next, in Sec. 3 the immersed interface method is briefly outlined.The jump conditions for the velocity and pressure in the Stokes equations are developed in Sec. 4. The derivation of analytic test cases is shown in Sec. 5 while results and convergence analysis follow in Sec. 6.A brief conclusion and future work follows in Sec. 7. 2. Governing Equations.Let a three-dimensional bounded domain, Ω, contain a closed and incompressible material interface Γ.For simplicity assume that the interface is wholly contained in the domain.The region enclosed by the interface is denoted as Ω − while the region outside the interface is denoted by Ω + , Fig. 2.1.The fluid in each domain is modeled using the Stokes equations: with boundary conditions u + = u b on the boundary of Ω.The pressure, p, body force term, g, and viscosity, µ, are all piecewise constant in the domain, with a finite jump occurring across the interface.The velocity field, u = (u, v, w), is assumed to be C 0 −continuous across the interface.It is also assumed that the velocity obey a no-slip condition on the interface.It is thus possible to state that the velocity of the interface Γ is given by q = u + = u − .Application of the volume incompressibility constraint, Eq. (2.2), to the momentum equations, Eq. (2.1), leads to a pressure-Poisson equation in each domain, with a boundary condition on the outer domain of (2.4) where n b is the outward facing normal to the domain Ω. The incompressible interface requires that the velocity on the interface be divergencefree: (2.5) where ∇ s = (I − n ⊗ n) ∇ is the surface gradient and ∇ s • is the surface divergence operator. The presence of the interface results in a singular force, f , exerted on the surrounding fluids.This singular force is application specific.In vesicle/cell simulations this force would be composed of a variable surface tension, γ, and other singular forces, f 0 : where n is the outward facing normal to the interface Γ (from Ω − into Ω + ) and H is the total curvature.An example of other singular forces which might be applied are the bending forces in vesicle and red-blood cell simulations [19,20] while the surface tension will be chosen to enforce the surface incompressibility constraint, Eq. (2.5).Future work will consider singular forces of this form.The singular force is balanced by a jump in the normal component of the stress tensor: (2.7) where the stress tensor is given by (2.8) 3. Immersed Interface Method.The governing equations as described in Sec. 2 result in two coupled, but discontinuous, fluid fields.In this section the scheme developed in Ref. [17] and used to solve the two-dimensional Stokes equations in Ref. [6] is reviewed.This is also the impetus for deriving the normal jump conditions across the interface. Begin by discritizing the domain Ω using a uniform Cartesian mesh with a grid spacing of h in all directions.While this section will focus on two-dimensional domains the extension to three-dimensions is straight-forward and used in the subsequent sections.Now consider the solution to a generic Poisson problem of the form where both u and f are (possibly) discontinuous across the interface.Upon discritizing Eq. (3.1) it is possible to classify all grid points as either regular or irregular, Fig. 3.1.Regular nodes are those nodes where the interface does not cross the discritized stencil.As the solution is continuous away from the interface the discretization at regular nodes does not need to be modified.Irregular nodes are defined as those where the interface does cross the discritized stencil.As the solution u may be discontinuous across the interface it is not possible to use the standard discretization.Instead the system is modified at irregular points to take into account the discontinuity. G f x i,j Consider the irregular point x i,j shown in Fig. 3.1.Let this point exist in the Ω + domain.A standard second-order finite difference discretization results in The issue is the values at x i−1,j and x i,j−1 exist in the Ω − domain, not the Ω + domain.This fact must be taken into account in the discretization.Let a jump across the interface be defined as where x Γ is a point on the interface and n is the outward normal vector.Assume that the jumps in the solution, [u] Γ , the first normal derivative, [∂u/∂n] Γ , and the second normal derivative [∂ 2 u/∂n 2 ] Γ are known on the interface for the problem given in Eq. (3.1).Using a Taylor Series expansion in the normal direction about an interface point the jumps may be extended to a grid point by where φ i,j is the signed-distance from the grid point x i,j to the interface, Fig. 3.1.Note that by using the signed distance functions it is possible to account for extension of the jumps into either domain.By extending the solution jumps from the interface to the grid points it can be written that u + i,j = u − i,j + [u] i,j .Use this expression in the Eq.(3.2) to get As the jumps [u] i−1,j and [u] i,j−1 are known they can be moved to the right-hand side as explicit corrections on the linear system, where C i,j is the total correction needed at an irregular node at location x i,j .In This method is easily extended to irregular nodes on either side of the interface and to three-dimensional systems. It should be noted that the extension of the jumps must be calculated up to third-order accuracy to ensure that irregular nodes have a local truncation error of O(h).Despite this higher local truncation error the overall method will retain the second-order accuracy of the underlying discretization [1,8,10].If the second-normal derivative is not taken into account the local truncation error for irregular nodes will be reduced to O(1) and the overall scheme would only be first-order. where the corrections C u and C p are non-zero only at irregular grid points.To calculate these corrections at irregular grid points the jumps up to the second-normal derivatives need to be derived for both the pressure and the augmented velocity, µu.This section will outline the derivation of these jump conditions.It should be noted that a C 0 -continuous velocity indicates that [u] = 0.It should also be noted that the interface can be approached from either the Ω − or Ω + domains.It is thus defined that where f is a quantity, such as velocity, of interest. Let n be the outward (pointing into Ω + ) unit normal while t and b are two unit tangent vectors chosen such that (n, t, b) form a Darboux Frame.Also chose t and b to lie along the principle directions of the interface, Fig. 4.1.Denote the directions in the normal and two tangent vectors as n, t, and b, respectively.In the following sections all derivative in a particular direction are denoted as ∂/∂n, ∂/∂t, and ∂/∂b.Note that a lack of bold-face indicates a direction and not a vector.Theorems 4.1 to 4.3 are non-trivial and non-obvious relations that will be used to develop the jump conditions for the pressure and velocity.Theorem 4.1.The jump in the viscosity times the normal derivative of the velocity in the normal direction is zero: Proof.For the fluid in each domain the incompressibility conditions holds, µ ± ∇ • u ± = 0. Thus, the jump in the incompressibility condition is zero across the interface.In terms of the local vectors this is The velocity as you approach the interface must also be surface-divergent free, µ ± ∇ s • u ± = 0. Therefore the the jump in the surface divergence of the velocity must also be zero, As the normal vector is continuous across the interface, n − = n + , the jump condition, Eq. (4.5), is obtained.Theorem 4.2.The following expression holds on the interface, Proof.The incompressibility constraint, ∇ • u = 0, must hold identically in both domains.Thus, the normal derivative of this constraint must also be zero, where the fact that [∂n/∂n] = 0, [∂t/∂n] = 0, and [∂b/∂n] = 0 are held to be true. Theorem 4.3.The following expression holds on the interface, Proof.The full Laplacian in the local frame can be written as (4.12) while the surface Laplacian can be written as [26] (4.13) Thus the following holds, (4.14) Using the result of Theorem 4.1 the expression in Eq. (4.11) is shown to hold true. The next three theorems outline the derivation of the pressure jump conditions.Theorem 4.4.The jump in the pressure is given by Proof.The balance of forces on the interface, Eq. (2.7), is (4.16) Taking the normal component of this balance results in (4.17) Using Eq. (4.5) results in the jump of the pressure field, Eq. (4.15). Theorem 4.5.The jump in the normal derivative of the pressure may be written as or alternatively as where κ t and κ b are the principle curvatures along the t and b directions and recalling that q is the velocity on the interface. Proof.As Eq. (2.1) holds in each domain the jump in the system must be zero, [∇p] = µ∇ 2 u + [g].Dotting this by n results in Eq. (4.18).As has been noted elsewhere [5] this form is not useful for numerical simulations due to the second-order partial derivatives of the velocity. To obtain the alternative form begin by taking the surface divergence of the force balance on the interface, Noting that the jump operator commutes with differentiation along the interface [5,28], individual portions of Eq. (4.20) can be written as Next, write the surface divergence of a vector v on the interface as Then, (4.23) When considering the derivatives of ∂u/∂n = (∇u) • n along the tangent directions it is necessary to also take derivatives of the normal vector along the same directions.This results in the following expression along the t-direction where the derivative along the b-direction is similar.In the Darboux Frame it can be written that ∂n/dt = κ t t + τ t b, where τ t is the geodesic torsion along the t direction.As the t-direction is a principle direction then the geodesic torsion is zero, τ t = 0 [3].Using this results in the following expression, Combining this with a similar result in the b-direction gives (4.26) In a similar fashion the final portion of Eq. (4.20) can be written as Next, use Eq. ( 4.12) to split the Laplacian term in the base jump condition, Eq. (4.18), and combine with Eq. 4.28 to obtain (4.29) Letting κ g t be the geodesic curvature along the t direction allows for the following to hold, When Eq. (4.31) is dotted by the unit normal vector and noting that [∂u/∂n]•n = 0, see Ref. [23], the expression ∂ 2 u/∂t 2 • n = 0 holds.A similar expression holds in the b direction. The exact form given in Eq. ( 4. 19) can be obtained by writing jumps of the form . This allows for the following to hold, Replacing the appropriate terms in Eq. (4.29) results in the alternative jump in the normal pressure derivative. Theorem 4.6.The jump in the second-normal derivative of the pressure is given by where the jump in the first-normal derivative of the pressure is given in Eq. (4.19). Proof.As the pressure-Poisson equation, Eq. (2.3), holds in each domain the jump is given by ∇ 2 p = [∇ • g].Splitting the Laplacian of the pressure into the local frame results in (4.34) Solving for the jump in the second derivative and using the result of Theorem 4.4 results in Eq. (4.33). The final three theorems outline the derivation of the velocity jump conditions.Theorem 4.7.The jump in the augmented velocity is given by Proof.Expanding the jump [µu] results in due to [u] = 0 and u + = u − = q on the interface. Theorem 4.8.The jump in the normal derivative of the augmented velocity is given by where P is the projection operator given by P = I − n ⊗ n. Proof.Dot the force balance on the interface, Eq. (2.7), by the tangent directions to obtain In conjunction with Eq. (4.5) a linear system can be written as , As the inverse of the matrix in Eq. (4.40) is the transpose of the matrix the jump can be obtained as Realizing that the term (f • t) t + (f • b) b is simply the tangential projection of the force f onto the interface the jump given in Eq. (4.37) is obtained.Theorem 4.9.The jump in the second normal derivative of the augmented velocity is given by Proof.Consider the jump of the augmented momentum balance equation, Eq. (4.1).Expand the Laplacian of the velocity and the gradient of the pressure into normal and tangential components: Due to continuity of tangential derivatives, the unit normal, and the total curvature, it can be stated that The jump in the pressure is simply Finally, noticing that (∂/∂t)t + (∂/∂b)b = ∇ s is the surface gradient the jump condition given in Eq. (4.42) can be obtained. Sample Test Case. To verify the accuracy of the derived jump conditions a sample analytic test case has been created.To create this test case valid inner and outer velocity fields, u − and u + , are determined for a pre-determined interface.These velocity fields are chosen to enforce the conditions u − = u + and ∇ s •u − = ∇ s •u + = 0 on the interface Γ and ∇ • u − = ∇ • u + = 0 in the domain Ω.Next, arbitrary inner and outer pressure fields, p − and p + are chosen.Using the velocity and pressure field the inner and outer body forces, g − and g + can be calculated using Eq.(2.1).The singular force, f is obtained through the force balance on the interface, Eq. (2.7). The analytic body and singular forces are then used in conjunction with the known interface velocity q and appropriate boundary conditions to solve the discontinuous Stokes equations in the domain using the Immersed Interface Method.The numerical pressure and velocity fields are compared to their analytic versions to determine the overall spatial accuracy of the method. In this example the interface is a unit sphere centered at the origin.The outer viscosity is taken to be 1 while the inner viscosity is given by µ.The exact inner velocity and pressure for the region x 2 + y 2 + z 2 ≤ 1 are given by The exact outer velocity and pressure for the region x 2 + y 2 + z 2 > 1 is p + =0, (5.5) which results in a body force of (5.6) A force balance on the interface results in a singular force of (5.7) Clearly on the interface q = u − = u + holds on Γ and can be applied in the derived jump conditions.6. Results.In the following results the domain is a taken to be a [−2, 2] 3 cube.The pressure and velocity fields are obtained using a split calculation where the pressure is calculated first by solving Eq. (2.3) and then obtaining the velocity field through the solution of Eq. (2.1).It should be noted that the jump conditions allow for any Stokes field solution technique, not only the one used here. Sample pressure results for the z = 0 plane on a 129 3 grid are shown in Fig. (6.1).The external viscosities shown are 0.001 and 1000.It is clear that the sharp discontinuity of the pressure is captured by the method and the result is not greatly affected by the viscosity ratio.To demonstrate that the derived jump conditions, when used with a secondorder Immersed Interface Method, achieve the desired accuracy the errors for both the pressure and velocity fields are now presented.In calculating the error external viscosities ranging from 10 −3 to 10 3 have been considered.Grid sizes ranging from 48 3 to 384 3 have been used, corresponding to grid spacings of h ≈ 0.0115 to h ≈ 0.0851.The error for the pressure field are presented in Fig. 6.2 while the corresponding velocity errors are shown in Fig. 6.3.For both the pressure and velocity fields the second-order accuracy of the underlying discretization is clearly preserved despite the presence of the discontinuities in the solution field.It should be noted that the overall accuracy of the method depends on the viscosity difference between the inner and outer fluid, with matched viscosity producing the lowest error.This is to be expected as the jump conditions are heavily dependent on this viscosity difference.Discretization errors in the surface derivative quantities, such as the surface Laplacian of Eq. (4.19), are amplified as the viscosity difference increases.While more accurate surface derivative calculations would reduce this dependence, it will always remain and must be taken into account. 7. Concluding Remarks.In this paper the three-dimensional, normal jump conditions for the pressure and velocity fields have been derived for an embedded incompressible/inextensible interface with a singular force and in the presence of a discontinuous viscosity field.The jump conditions take into account the additional constraint of interface incompressibility given by a surface-divergent free velocity field.The jump conditions are applicable to any Stokes solver using the Immersed Interface Method.As a demonstration of the accuracy of the jump conditions a sample analytic test case has been created.Error analysis using this test case demonstrates that a second-order accurate discretization maintains the expected accuracy despite the discontinuity of the solution fields.Future work will use these jump conditions to construct generalized Immersed Interface Methods which are capable of enforcing surface incompressibility while solve the Stokes equations for multiphase Stokes flow with discontinuous viscosity. Fig. 3 . 1 . Fig. 3.1.Discritized Laplacian at a regular node (circle) and irregular node (square).Irregular nodes are those where the interface intersects the stencil.The signed distance function from any grid point to the interface is given by φ. 4 . Derivation of Jump Conditions.The governing equations in Sec. 2 result in two coupled Poisson problems.Using the Immersed Interface Method of Sec. 3 the two systems can be written over the entire domain by including the appropriate corrections: Fig. 4 . 1 . Fig. 4.1.The local vectors on an interface point.The unit vector n is the outward facing normal while t and b are unit tangent vectors which lie along the principle directions at the point on the interface. and b = (b x , b y , b z ) T are the components of the local vectors. Fig. 6 . 1 . Fig. 6.1.Example of the pressure distribution on the z = 0 plane for a 129 3 grid for outer viscositoes of 0.001 and 1000.The inner viscosity is equal to 1.	2013-09-04T20:48:00.000Z	2013-09-04T00:00:00.000	{ "year": 2013, "sha1": "01bc2a7e1e341a2b0daf9d7cf166a1d36d158a05", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "fa3255d9823aecdca85fb7720e1b15ffebda9318", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Physics", "Mathematics" ] }
236451065	pes2o/s2orc	v3-fos-license	Dimeric allostery mechanism of the plant circadian clock photoreceptor ZEITLUPE In Arabidopsis thaliana, the Light-Oxygen-Voltage (LOV) domain containing protein ZEITLUPE (ZTL) integrates light quality, intensity, and duration into regulation of the circadian clock. Recent structural and biochemical studies of ZTL indicate that the protein diverges from other members of the LOV superfamily in its allosteric mechanism, and that the divergent allosteric mechanism hinges upon conservation of two signaling residues G46 and V48 that alter dynamic motions of a Gln residue implicated in signal transduction in all LOV proteins. Here, we delineate the allosteric mechanism of ZTL via an integrated computational approach that employs atomistic simulations of wild type and allosteric variants of ZTL in the functional dark and light states, together with Markov state and supervised machine learning classification models. This approach has unveiled key factors of the ZTL allosteric mechanisms, and identified specific interactions and residues implicated in functional allosteric changes. The final results reveal atomic level insights into allosteric mechanisms of ZTL function that operate via a non-trivial combination of population-shift and dynamics-driven allosteric pathways. Introduction Light-Oxygen-Voltage (LOV) domains are widespread in nature, where they couple sensing of UVA/Blue light into regulation of diverse modes of biological activity. [1][2][3] Central to their function is an innate ability to couple cofactor chemistry to dynamic allosteric changes in protein conformation to regulate either protein-protein interactions, or activity of signal-transduction domains N-or C-terminal to the LOV domain. [2] The ability of LOV domains to regulate diverse signaling elements has led to their widespread use in optogenetic tools, [4][5][6] and has identified them as model proteins for studying allostery, the change of the protein structural conformation or states distribution due to a perturbation of a non-functional secondary site. [7][8][9][10] Structural and computational approaches have been used to study LOV signal transduction in a number of systems. [11][12][13][14][15][16][17][18][19][20] These approaches have identified a mechanism of signal transduction dependent on three conserved elements: i) UVA/Blue-light triggers formation of a covalent adduct between a flavin cofactor (Riboflavin/FMN/FAD), and a conserved Cysteine residue resulting in sp 3 hybridization of the flavin C4a position and protonation of the flavin N5 position [2,15]; ii) N5 protonation causes a conserved Gln residue, coplanar to the isoalloxazine ring (buried conformation), to flip to optimize H-bonding interactions to N5, and alter H-bonding interactions with elements in the Aβ-strand [11,12]; iii) The Gln flip and altered Hbonding interactions induce conformational changes within a central five-stranded β-sheet and/or N/C-terminal extensions to the core LOV domain (Ncap/Ccap) [13,16]. The resulting consensus mechanism indicates that the conserved Gln residue is indispensable for LOV-signaling, and N5 protonation is both necessary and sufficient for LOV allostery. [11,21,22] Recent structural and biochemical studies of the Arabidopsis thaliana circadian clock photoreceptor ZEITLUPE (ZTL) have called the consensus LOV signaling mechanism into question, and have suggested that LOV allostery may be more fluid across the LOV superfamily. [16,23] Structures of ZTL proteins in the dark-and light-states indicate divergent signaling elements compared to other LOV proteins: i) The conserved Gln residue (Gln154) adopts a dynamic and heterogeneous population rotating between buried and exposed conformations in the dark. Subsequent photoactivation biases the Gln population towards the buried conformation seen in all other LOV-structures to induce downstream signal transduction [16,23]. ii) Q154L variants are tolerated in ZTL and are naturally present in ZTL family members [23]. iii) Evolutionary analysis revealed that the altered signaling pathway was dependent on conservation of two residues in the Aβ-strand, Gly46 and Val48, that are co-conserved in all ZTL proteins [16]. Further, examination of other LOV structures with Gly residues at the position equivalent to Gly46 revealed altered Gln dynamics in these proteins, thereby highlighting conservation of allosteric motions linking the C-terminal Gln, to N-terminal conformational changes dependent on the residue identity at position 46 in ZTL [23][24][25]. Examination of G46S and G46A structures revealed coupling between C-terminal Gln dynamics, the residue identity at position 46, and global conformational changes. [16,23] G46S and G46A structures mimic a light-state like orientation of the active site Gln that is coupled to the movement of a conserved Phe residue (Phe156), C-terminal salt-bridge formation, and the movement of the Ncap. Specifically, rotation of Gln154 towards the buried conformation disrupts contacts to the N/Ccap to induce a 180˚rotation about the LOV-dimer interface ( Fig 1). However, these observations relied on static structures of allosteric variants that may block observation of dynamic motions necessary for light/dark regulation of ZTL structure, and may not necessarily reflect allostery in WT proteins. Further, the use of allosteric variants to trap specific ZTL configurations only allows for snapshots of starting and end points in an allosteric process, and not the dynamic motions gating the altered mechanisms of signal transduction, thereby limiting understanding of how the ZTL allosteric landscape diverges from other LOV proteins. To surpass this limitation and obtain a dynamical description of ZTL allostery, exploring the conformations that lie in between the starting and end points in the allosteric process to ultimately extrapolate the structural changes during this process is necessary. In recent years, computational approaches to advance a quantitative characterization of allosteric mechanism in proteins have been developed. Molecular dynamics (MD) simulations are a useful tool to explore the conformational landscape of a protein by providing the evolution of a system over time at an atomistic level. [26][27][28] Combined with MD simulations, Markov state model (MSM) approaches can provide connectivity maps of states on the free energy landscape, estimate the effect of allosteric perturbations on the conformational equilibrium, and rigorously describe kinetics of allosteric transitions. [19,[29][30][31][32][33][34] Herein, we employ atomistic simulations that integrate Markov State modeling, and supervised machine learning classification models to elucidate a dynamic allosteric process gating ZTL function. These results confirm a complex and dynamic ZTL landscape that links sequence motifs within the Aβ and Iβ strands to an insertion between the E/F helices (E-F loop), which drives the reorientation of the ZTL LOV-dimer. The results provide keen insight into the role of Gly46 and Val48 mutations in disrupting ZTL allostery that have implications for plant physiology, and broader impacts to LOV allostery. The ensemble of computational tools employed in this study enabled us to validate and integrate otherwise inaccessible information to the experimental investigations of ZTL allostery. The MD simulations support the stability of dimer complexes, and also reveal the propagation of the allosteric perturbation starting from the flavin photoreceptor to the overall protein structure. The remainder of the paper is organized as follows. First, it is presented how Gln154 responds to the change in state of the flavin and the impact of G46S and V48I variants on Gln154 dynamics. Then how the different conformations that Gln154 adopts impact the interactions between residues in the functional N-and C-termini are investigated. Several intramolecular interactions that could be crucial to propagate the signal throughout the protein are also investigated. Particularly, how these intramolecular interactions affect the population of functional metastable states on the conformational landscape of the protein is evaluated. Towards these goals we adopted the dimensionality reduction technique tICA and Markov state modeling. Lastly, with the use of Machine Learning we elucidated characteristic structural changes that correlate with different metastable states involved in ZTL allosteric process. Results To elucidate a model of ZTL allostery, we performed a series of MD simulations on ZTL structures for either wild-type (WT) or allosteric variant forms of the isolated LOV domain. Specifically, we simulated the WT dark-state in an anti-parallel conformation (PDB ID: 5SVG), the light and dark anti-parallel conformations of the V48I:G80R variant that disrupts light-driven allostery (PDB ID 5SVV and 5SVW, respectively), WT and G46S:G80R (PDB ID 6WLP) variants in a parallel conformation that mimic the light-state conformation of Q154, and transient structures for both the anti-parallel and parallel conformations. All systems subjected to the simulations in this study are listed in Table 1. Initial focus was placed on resolving several important limitations of allosteric variant structures. Namely, we anticipated to verify two key aspects of ZTL signaling independent of allosteric variants or crystal contact restraints: 1) Verify stability of the ZTL LOV parallel dimer in WT light-state proteins (Protein Stability Analysis in S1 Text); 2) Examine Q154 conformational states and dynamics in WT and allosteric variants of ZTL LOV dimers. Role of Gln154 in the function of ZTL dimer ZTL function within the circadian clock is dictated by daily changes in light intensity. The consensus model of LOV signaling involves a light-driven Gln-flip in response to changes in flavin N5 protonation. [11,16,17,24] In most LOV structures, the conserved Gln residue occupies a buried conformation coplanar to the flavin isoalloxazine ring. In the dark state, N5 is unprotonated, with the active site Gln forming H-bonds to the N5 and O4 positions via its amide functionality. [16,17,24] Upon illumination, N5 becomes protonated, leading to 180˚rotation of the active site Gln to allow H-bond formation between the Gln-carbonyl group and N5-H. [16,17,24] The result is a disruption of contacts with residues in Aβ strand promoting conformational changes in N/C-terminal extensions to the LOV core. [11,16] Structures of ZTL indicate an alternative mechanism, where in the dark state Gln154 is dynamically adopting heterogeneous conformations between exposed (perpendicular to the isoalloxazine ring) and buried conformations. Light-activation was then predicted to drive rotation of Gln154 towards the buried conformation leading to downstream signal transduction. [16] To characterize the protein structural and dynamical profile related to the Gln154 switch, we focus on the analysis of Gln154 orientations corresponding to different FMN states in WT and allosteric variants. Based on the hydrogen bonds between Gln154 and FMN, we identified three main conformations: exposed, buried-I, and buried-II. The exposed conformation is identified when the only interaction presented between the FMN and Gln154 is a hydrogen bond between the side chain amino group and the O4 position of the FMN (Fig 2B). The buried-I conformation is formed when a hydrogen bond between the Gln154 side chain amino group and the N5 of the FMN is present (Fig 2C), in addition to the hydrogen bond in the exposed conformation. The buried-II conformation is formed when the carbonyl moiety of the Gln154 side chain forms a hydrogen bond with the protonated N5 of the FMN (Fig 2D). To quantify which conformations of Gln154 are sampled in different ZTL structures and states, we used a combination of the chi1, chi2, and chi3 dihedral angles (S4 Fig), which can uniquely identify different Gln154 conformations. The representative values for each conformation are provided in S1 Table. To quantify which conformations Gln154 explores in different ZTL structures and states, we performed a similarity analysis. For each frame, the set of dihedrals are compared to the representative dihedrals of the Gln154 conformation described above. The conformation is assigned based on similarity (Fig 2G and 2H). Consistent with the WT structures, these three conformations differentiate dark-and light-state configurations of WT ZTL. Specifically, in the dark state, Gln154 is heterogeneous and samples both exposed and buried-I conformations with a preference for the buried-I conformation (Fig 2B and 2C). Photoactivation leads to a strong bias toward the buried-II conformation observed in the light-state ZTL and other LOV structures, supporting the ordering of Gln154 locus. In contrast to WT MD simulations, exposed, buried-I, and buried-II conformations are insufficient to describe the conformational states of Gln154 in allosteric variants G46S:G80R and V48I:G80R. Specifically, in V48I:G80R a significantly exposed conformation (chi2 angle of approximately -177˚; Exposed-II conformation) with an additional stabilizing H-bond formed with Asn123 is the predominant orientation (Figs 2E and S5). The exposed-II conformation is consistent with V48I:G80R structures, which are trapped in a predominantly darkstate conformation and functions as dominant-dark proteins in vivo. [16] In G46S:G80R, a different additional conformation is present, in which the amino group of the Gln154 side-chain forms a H-bond with the O4 position of FMN in addition to the H-bond in the buried-II conformation ( Fig 2F). We refer to this latter state as an extended conformation. The Gln conformational analysis provided insight into both the ZTL mechanism, and variances between Gln dynamics in MD simulations and the static structures. First, as noted above, dihedral-based conformation classification demonstrates a dynamic dark-state Gln154 conformation that flips between buried-I and exposed conformations. A dynamic dark-state Gln conformation is consistent with WT ZTL structures. [16,23] Light-activation leads to Gln154 adopting a buried-II conformation for over 80% of the trajectories (Fig 2H), confirming light-driven ordering of Gln154 in a buried conformation. In contrast to WT proteins, G46S:G80R Gln conformational dynamics deviates from those observed in static crystal structures. Experimental studies showed that in G46S:G80R and G46A:G80R structures Gln154 largely occupied a single conformation where the Gln154 side chain was rotated toward the buried-I conformation. In contrast, MD simulations indicated PLOS COMPUTATIONAL BIOLOGY that G46S:G80R could sample the exposed conformation, and could further sample a unique extended conformation containing H-bonds to both the N5 and O4 positions. These divergent observations can provide some insight into prior in vitro functional assays of G46S containing variants. Even though G46S variants demonstrate light-state like activity, exposure to light can still enhance light-driven complex formation with GIGANTEA. [23] These aspects can be explained by the Gln conformational analysis, specifically, the ability to form both Buried-II and Extended conformations. Both conformations result in Gln154 being largely co-planar with FMN, and prevent sampling of dark-state like exposed conformation leading to an enhanced light-state response. Overall, these results complicate the Gln-flip mechanism. Although ZTL and LOV proteins retain a light-state signaling buried-II conformation, ZTL light-state activation largely results in ordering of the Gln154 side chain as opposed to a simple Gln-flip mechanism observed in other LOV proteins. Further, side chain identities at the G46 and V48 positions in Aβ strand can significantly perturb the Gln154 landscape to disrupt allosteric signal transduction. Although providing insight into the proximal light-driven signaling events, these observations lead to two conundrums regarding reconciling aspects of downstream signal transduction in ZTL: i) If Gln154 H-bonds are still driving Gln154 ordering and signal transduction, how do ZTL proteins signal in a Q154L background; ii) If G46S proteins are still sampling an exposed Gln conformation, why do they prefer the light-state parallel LOV dimer. To answer these two questions, we performed detailed analyses of Gln conformational effects on N/C-termini conformational changes dictating dimer reorganization and light/dark-functionality. Gln154 dynamics regulates downstream allosteric changes C-terminal Conformational Changes and Gln154. Structural studies of ZTL allosteric variants identified two structural motifs coupled to Gln154 conformational dynamics. These consist of an N-terminal Cys45-Gly46-Phe47 (CGF motif) and a C-terminal Gln154-Phe155-Phe156 (QFF motif) that are predicted to couple Gln154 conformational changes to the Ncap and Ccap, respectively. In allosteric variants that trap a light-state-like conformation (G46S:G80R), rotation of Gln154 to the buried conformation is coupled to the movement of Phe156 towards the flavin binding pocket, and formation of a salt-bridge between Glu158 and Arg125. [23] Analogous conformational changes have been observed in LOV structures that contain a Gly residue at the position equivalent to Gly46, where a different conserved Phe residue (Phe66 in ZTL) undergoes light driven conformational changes coupled to Gln reorientation. [23,24,35] To identify the role of Phe residues in dictating signal transduction in ZTL, we examined Phe66 and Phe156 conformations in WT and variant ZTL trajectories. Consistent with structures of ZTL allosteric variants, we observe that Gln154 rotation from exposed to buried-I and/or buried-II conformations is coupled to ordering (pulling) of the terminal part of Iβ strand and movement of Phe156 towards a cleft vacated by Gln154 rotation. The pull-effect has been quantified by monitoring the distance between the Cα of Phe156 and the O4 of FMN ( Fig 3A). Notably, the pull-effect is enhanced in WT light-state structures where Phe156 lies closer to the active site flavin (Fig 3B and 3C). Examination of the allosteric variants V48I:G80R and G46S:G80R confirms alteration in Phe156 movement, where V48I negates, and G46S enhances the pull-effect, consistent with dominant-dark and dominantlight state signaling respectively. These results verify that Phe156 movement towards the active site flavin is a proximal event dictating C-terminal conformational changes following adduct formation, and a determining factor in allosteric ZTL variants. Previous structural studies suggested that movement of Phe156 may be stabilized by a πstacking interaction between Phe66 and Phe156. [16,23] Further, the residue equivalent to Phe66 undergoes light-driven conformational changes in LOV structures containing a Gly at position 46. [24,35] To further understand the role of Phe residues in LOV signal transduction, we examined Phe66 dynamics in WT and variant ZTL structures. The concerted movement of Phe156 and Gln154 rotation was assessed and quantified using the Pearson Correlation (PC) coefficient (S2 Table). In ZTL, we observe significant movements of Phe156 and Phe66 towards the FMN binding site that is coupled to Gln154 rotation. Movement of Phe66 mirrors analogous light-driven conformational changes in Bacillus subtilis (YtvA) and a short LOV protein in Pseudomonas putida (PpLOV) that contain Gly residues at positions equivalent to Gly46. [24,35] In ZTL, movement of Phe66 and Phe156 as a consequence of Gln154 rotation produces a π-stacking interaction that is light-state selective (Fig 4B and 4C), where we observe a narrow spread and higher density of the distributions in the light-state structures. Examination of allosteric variants V48I:G80R and G46S:G80R demonstrates perturbation of Phe156 and Phe66 movement. As expected, introduction of V48I sterically hinders Gln154 rotation (S6A Fig) and prevents Phe156 movement, consistent with locking ZTL into a darkfunctional state. Because the dark-state of G46S:G80R mutant mimics the light state conformation, it is expected that this state also favors the light-state-like π-stacking interaction between Phe66 and Phe156. However, this is not the case as the dark-state of G46S:G80R mutant displays the lowest peak of the Phe66 and Phe156 distance distribution among all dark states ( Fig 4B). According to the pull of the Phe156 (Fig 3) and the consequent salt bridge at the C-terminal (Fig 5), the light-state-like allosteric perturbation is still present in the G46S:G80R mutant. Since the π-stacking between Phe66 and Phe156 seems not to be a key factor to stabilize Phe156 in the pocket, an alternative interaction should take place for this purpose. Examination of distances between Gln154 NE2 and Phe156 CZ highlights an alternative allosteric route coupling Gln154 and Phe156 movement. A Coulomb interaction between the partial positive charge of the amino group and the negative charge of the aromatic ring couples Gln154 and Phe156 resulting in characteristic short distance peaks in both dark-and light-states of G46S: G80R (Fig 4E and 4F). In this manner, electrostatic interactions drive a light-state like "in" conformation of Phe156 regardless of light-driven π-stacking interaction with Phe66. Crystal structures of the allosteric variant G46S:G80R indicate that Phe156 movement may be coupled to Arg125-Glu158 salt bridge formation. [16,23] However, whether this salt-bridge formation was an artifact of crystal packing interactions, or mainly due to the parallel dimer orientation of G46S:G80R, or unique to the G46S allosteric variant was unknown. To further reveal the light-driven conformational changes between the proximal CGF motif and the Ccap, we analyzed the correlation between FMN light-states and formation of the Arg125-Glu158 salt bridge in all WT and variant trajectories. In the WT dark-state, as well as both dark-and light-states V48I:G80R trajectories, the C-termini were free to fluctuate and exhibited Arg125-Glu158 distances incompatible with salt-bridge formation ( Fig 5B). In contrast, WT light-state, as well as both dark-and light-states G46S:G80R, trajectories exhibited a sharp peak around 4 Å consistent with a stable salt bridge between these two residues ( Fig 5C). The distance analysis reveals the coupling of conformational changes in the QFF motif to longrange conformational changes following adduct formation in WT proteins. These confirm that the effects of allosteric variants at the G46 (light-state like) and V48 position (dark-state like) propagate to signaling elements within the Ccap. Overall, analysis of putative signaling elements within the proximal QFF motif and downstream Ccap elements suggests a concerted mechanism coupling Gln154 dynamics to ordering of the C-termini. For all variants and light-states, salt-bridge formation directly correlates with adduct formation, Gln154 ordering, and movement of Phe66 and Phe156 to form stabilizing π-stacking interactions. These conformational changes are dependent upon the unique presence of a Phe residue at position 156, and delineates a mechanistic route at an atomistic level. N-terminal Conformational Changes and Gln154. In addition to conformational changes within the Ccap, critical structural changes for light-dependent ZTL function occur at the Ncap. Original models of ZTL signaling relied on a shift in population of Gln154 from a heterogenous, but predominantly exposed, conformation in the dark state to an ordered buried conformation in the light state. [16] Although the above analysis shows a heterogeneous population in the dark, the buried-I conformation still dominates, and both the buried-I and buried-II conformations show the same degree of rotation around the chi2 angle (S1 Table). Thus, Gln154 orientation alone appears to be a poor choice of driving structural change at the Ncap. Further, given that G46S variants can still sample both buried-I and exposed conformations, what factors dictate the observed light-state like allosteric effect at the Ncap are poorly understood. To alleviate these discrepancies, we examined dark-and light-state trajectories to identify additional interactions that differentiate buried-I, buried-II, and exposed conformations. Our analysis indicates that in the buried-II conformation, the amino group of the Gln154 side-chain can form an H-bond with the proximal Gly46 carbonyl moiety, which is located at the interface between Aβ and the Ncap (Fig 6A). This interaction results in a "pulling" effect on Gly46 specific to the light-state buried-II conformation (Fig 6B and 6C). The "pulling" effect observed in WT ZTL structures is analogous to allosteric mechanisms in the fungal PLOS COMPUTATIONAL BIOLOGY circadian clock protein VVD, [11] where the Gln-flip (buried-I to buried-II) permits formation of an H-bond between the active site Gln and the carbonyl of Ala72 and pulling of the Ncap towards the LOV core. Importantly, Ala72 structurally occupies the same site in Aβ strand, indicating conservation of a signaling motif in fungal and plant circadian clock proteins. Notably, V48I:G80R and G46S:G80R allosteric variants are exceptions to this mechanism. The first is expected, as it prevents the Gln154 from switching to a buried conformation via steric hindrance that functionally locks ZTL in a dark-state-like conformation. For the latter, a more in-depth investigation is necessary to explain how this mutant is able to cause a lightstate functional perturbation within the Ncap. Examination of G46S:G80R structures and trajectories identifies an alternative allosteric route, in which the steric presence of the Ser46 side chain alters the distance distribution between the Ncap and Gln154 leading to an increase in the distance of Ser46 from the LOV core ( Fig 6D). From these structural analyses, we delineate two alternative routes for propagation of allosteric effect following light-activation. We observed that in the WT, light activation first results in ordering of Gln154 in a buried-II conformation, leading to the formation of: i) π-stacking interactions between Phe66 and Phe156, which is responsible for propagating the structural change to the Ccap, ii) H-bond between Gln154 and Gly46, which is responsible for propagating conformational changes to the N-termini. This WT allosteric mechanism is altered in the G46S:G80R variant where the presence of Ser46 induces: i) a unique extended Gln154 conformation, which stabilizes Phe156 in the FMN binding pocket via electronic interactions resulting in propagating a structural change to the Ccap, and ii) a Gln154 conformation independent, steric-based conformational change at the ZTL N-termini. Gln154 effect on ZTL conformational flexibility The investigation of the structural changes in the N-and C-termini granted only a partial picture of the possible downstream effects of ZTL light-dependent Gln154 conformational gating. In fact, shifts in the population distributions on the conformational landscape without appreciable structural transformations can contribute to the allosteric protein functional changes. [36][37][38][39][40][41][42][43][44][45][46][47][48][49] To show how alteration in Gln154 conformational sampling in WT ZTL and allosteric variants impacts the ZTL structural landscape, we performed time-structure Independent Component analysis (tICA) by featurization of the protein conformation using all the Cα pairwise distances. The resulting 2-dimensional plots represent the impact of light-and dark-states and allosteric variants on the ZTL conformational space explored. The two components corresponding to the slowest-relaxing degrees of freedom, which associate with functionally relevant motions, distinguish the native dark-and light-state structures (Fig 7). Specifically, the native dark-state (Red, dark_wt_anti), samples considerably more conformational space than the native light-state (Maroon, light_wt_para), consistent with ordering following adduct formation (Fig 7A). To verify the effect of adduct formation and Gln154 conformation on ZTL dynamics, we further analyzed the conformational space explored in the transient structures, which mimics the initial landscape sampling immediately following adduct formation (transient-light) and scission (transient-dark). Consistent with disorder-order transitions, the transient dark state (tr_dark_wt_anti) stretches from the light state region to the dark native region. Further, the transient light state (tr_light_wt_anti in Fig 7C) displays dynamical behavior similar to the light state. This is indicated by its overlapping with the native light state structures in the tICA plot and by significantly less conformational space explored. These results confirm that light drives a reversible disorder-order transition that is directly coupled to adduct formation and Gln154 dynamics (Fig 7). Examination of how the conformational landscape is perturbed by G46 and V48 allosteric variants provides insight into how subtle local alteration in side-chain identity can impact dynamics-driven allostery. Specifically, the conformational space sampled by the V48I:G80R mutant in the dark state is more restricted than the WT-dark state and largely overlaps with the anti-parallel transient states (dark_v48i_g80r in Fig 7). These results are consistent with restricting the Gln154 conformation to a primarily exposed conformation. Similarly, the light- state conformations are more restricted, leading to only minimal overlap between V48I:G80R and the native light parallel simulations (Fig 7C). Thus, introduction of V48I restricts dynamic motions gating interconversion between light-and dark-configurations. In contrast, G46S variants lock the dynamics away from the dark state structures (dark_g46s_g80r and light_g46s_g80r in Fig 7), supporting light-state-like functionality in the absence of the flavin C4a adduct. In summary, tICA was used to analyze the effect of mutations and light state on the overall protein dynamics. In particular, the dark state displays higher flexibility, while the light state seems to have limited protein motions. Although the tICA plots show overlap between parallel and anti-parallel structures, it is important to point out that in our simulations the inter-conversion between the parallel and anti-parallel dimers was not fully observed. Rather, this analysis captured dynamic similarities between different systems, which were strongly influenced by monomeric atomic motions (S7 Fig). We observed a correlation between the protein flexibility deduced by the tICA projections and the Gln154 conformational dynamics. This indicates the relationship between Gln154 and the protein dynamics, whereby an ability to interconvert between Gln154 conformations is essential to drive light-dependent allosteric conversions in ZTL structure. To better understand how these dynamic motions kinetically gate conformational changes we exploited the reduced dimensional space generated using the tICA to generate a Markov state model. Identification of functional stable states in ZTL conformational landscape Markov State Model (MSM) can help to discretize the protein conformational landscape into functional metastable states and obtain a kinetically relevant picture of ZTL allosteric process. To achieve this goal, we followed the following protocol: i) k-means micro-clustering into microstates, ii) building of a MSM, and iii) Perron-cluster cluster analysis (PCCA) for transition-based macro-clustering. The details for each step can be found in the Materials and Methods section. Different metastable states lie in different free energy basins illustrating the effectiveness of PCCA in separating kinetically separated states (S8B Fig). As shown in Fig 8, the probability of a structure to remain in its original state is higher than the probability for it to transition to other states. This indicates that each macrostate is in a minimum on the free energy surface, and that the kinetic barriers prevent the system from switching among macrostates frequently. The occupation of these macrostates from different ZTL simulations is plotted in S9A (Fig 8B and 8C). The plot of the transition probabilities among different metastable states (Fig 8) shows that the states contained in the transient area (States 3, 5, 7, 8 and 9) have higher probability to transit to the states occupied by the native conformations. This confirms the role of the transient states as intermediates between the stable native dark and native light states. When the FMN cofactor is modeled in the light state, the conformations of the protein shift from dark state-like State 2 to the intermediates between the State 2 and the light-state like State 5. This shift of conformations mimics the biological process of the protein changing its functional dynamics and structures upon change of light conditions. [13,15] As shown in Fig 8B, the native dark state structures cover a large area, exploring most metastable states except for States 4 and 5. This shows that the absence of the photo-induced covalent bond between FMN-Cys82 correlates with the high conformational flexibility. On the contrary, the distribution of the light states is limited in a narrow area (Fig 8C), suggesting that the photo-induced covalent bond between FMN-Cys82 correlates with a low conformational flexibility. Fig 8D illustrates representative structures for the 11 metastable states identified in the MSM. Interestingly, there is a partial rotation of the two monomers in State 11, suggesting a rotation of the monomers towards the anti-parallel conformation. This state is populated solely by parallel transition dark structures, whose RMSD was found to significantly fluctuate during the MD simulations (Protein Stability Analysis in S1 Text). The combination of these two observations could provide an ulterior dynamical confirmation of the light influence to ZTL dimer orientation. PLOS COMPUTATIONAL BIOLOGY However, from the visual inspection, we were not able to find other clear structural differences among different metastable states. For this reason, we employed a machine learning based classification model to identify other structural differences that may play a significant role in ZTL allostery. ZTL allosteric structural changes captured by Machine Learning Modeling In order to gain further structural insight into ZTL allostery, a machine learning (ML) model was employed to identify non-trivial structural changes among protein metastable states. This model is carried out on the entirety of the data generated using the MD simulations, and depicts a dynamical nature in the structural characterization of ZTL allosteric process. S10 Fig shows a validation of our method in discerning high importance structural features. Through this ML model, it is assumed that the structural features with high importance are likely coupled with free energy barriers dividing macrostates and thus provide an atomistic insight into the allosteric mechanism associated with the identified macrostates. In this ML model, the features constructed based on the protein structure, in this case all the Cα pairwise distances, are ranked using a percentage score based on their contribution in differentiating macrostates. Overall, the OvO-random forest classification model reached a validation accuracy 95% with 10 trees of 4 layers depth. 357 out of 33670 Cα pairwise distances account for 90% of the distinguishability among macrostates (S11 Fig). Structural effects of FMN states. Despite the fact that in our simulations we do not observe the complete interconversion between anti-parallel and parallel dimer conformations, it will be informative to find key conformational changes related to functional monomeric motions. To investigate structural changes not localized in the proximity of FMN binding site that arise in response to the changes in the FMN state, we applied ML analysis to key metastable states identified in the MSM. Of particular interest are the metastable states occupied mainly by the WT dark or light structures, and the transient states adjacent to them. The changes between the dark state and transient states were inspected by analyzing the most important structural differences that distinguish the State 2 as the dark native structure and State 3 representing a transient state adjacent to the State 2 ( Fig 8D). We then compared States 3 and 9, as a mean to compare transient metastable states adjacent to the native dark and light states, respectively. Finally, State 9 and State 4 as the light native structure were compared. Examination of these macrostates revealed coupling of key regulatory regions in ZTL. The ML classification model indicates that States 2 and 3 are distinguished from each other by undocking of the N-termini from the LOV-core and an increase in N-terminal rigidity (Figs 9B and S12B). The ML investigation of the structural differences between States 3 and 9 identifies the EF-loop as key to allosteric transitions. The EF loop is of unknown function and is only present in members of the ZTL-family and homologs of the fungal circadian clock photoreceptor VVD. In VVD, the EF-loop was presumed to be involved in accommodation of an FAD vs. FMN cofactor. However, such an explanation is not applicable to ZTL, which binds FMN. Rather, the EF-loop has been suggested to be involved in signaling by an unexplained mechanism. Here, ML classification models indicates that an increase in EF-loop rigidity is an essential structural feature to differentiate States 3 and 9 (Figs 9B and S12D). Examination of the light State 4 and adjacent transient State 9 indicates that the key structural feature that distinguishes these two states is an increase in rigidity of the N-termini, which moves closer to the LOV core (Figs 9B and S12F). Overall, in the structural characterization of ZTL allostery, changes in the structural features within the monomers are dominant (S13 Fig). Furthermore, an asymmetric change of the two monomers was identified among the key metastable states investigated (Fig 9A). This may allow for the increased sensitivity to light-intensities as the activation of one monomer may be sufficient to activate the allosteric process in an asymmetric manner. It is important to note that the overlap in the tICA space is due to the monomeric correlated motions of the anti-parallel and parallel dimers as the full transition between them was not observed (S7 Fig). Further sampling in the transition area between anti-parallel and parallel dimer conformations could reveal yet unexplored regions of the free-energy landscape that could reveal additional key conformational changes that are potentially involved in the monomers' re-orientation. Structural effect of ZTL variants. Comparison of the ML classification model with structural states occupied by allosteric variants provides further clarity on how mutations at the G46 and V48 positions alter dynamics driven allostery. First, introduction of V48I constrains Gln154 in an exposed conformation impacting dynamics at the N/C-termini, rigidifying these regulatory regions in dark-state conformations (State 7 ; Fig 8). Comparisons to features identified in the ML classification model indicates that rigidification of the N/C-termini is coupled to rigidification of the E-F loop (S14 and S16 Figs). Thus, V48I variants restrict Gln154 dynamics, leading to a decrease in dynamics at key structural features necessary to light-dark transitions. PLOS COMPUTATIONAL BIOLOGY Regarding the G46S mutation, the serine induces a unique Gln154 extended conformation, which enhances the light-state activity by preventing the exploration of the dark state exposed conformation. This mutant is shown to initiate the perturbation to the N-termini using an alternative mechanism where the Aβ strand is pushed further from the LOV core. The ML classification analysis identifies that as a result of this alternative route of perturbation, N-termini remain far from the dimer interface in this variant, while in the wild type it crawls into the dimer interface (S15 and S17 Figs). Overall, examination of the allosteric variants indicates two modes of allosteric perturbation. In V48I variants, the allosteric mechanism is not altered, rather V48I variants alter the dynamic dark-state landscape essential to ZTL allostery. Specifically, restriction of conformational freedom can trap the protein in dark-state-like configurations. In contrast, variants at the nearby G46 position can directly alter the allosteric mechanism to obtain structural changes in the functional N-termini as an effect of the steric hindrance caused by Ser46. Discussion and conclusion LOV proteins are ubiquitous in nature and have been employed in optogenetic tools to impart photodynamic control over signaling domains. Due to their widespread utility, LOV proteins have been a central focus in both structural and computational approaches to delineate allosteric mechanisms. Currently, the consensus mechanism in LOV signal transduction necessitates the presence of a Gln residue adjacent to the O4-N5 positions of the flavin cofactor, where changes in N5 protonation can drive a flip in the Gln side chain. Recent structural studies of WT ZTL and allosteric variants identified an alternative signaling mechanism that can persist in the absence of a Gln residue, whereby in WT proteins the Gln rotates from a heterogeneous exposed conformation in the dark to a buried light-state conformation. In these studies, it was proposed that conserved Phe residues may be able to sense electronic changes in the FMN cofactor and mediate a Gln-independent signaling mechanism. Notably, light-dependent movement of Phe residues has been observed in structural and computational studies of LOV proteins that contain Gly residues at the position equivalent to ZTL G46. Despite observation of analogous processes in other LOV proteins, whether the structures observed in allosteric variants represent the ZTL signaling mechanism remained unproven. Further, how Gln conformational dynamics alter allosteric responses remained unexplored. In the present study we have performed detailed atomistic simulations, MSM analyses, and ML classification models to delineate allosteric effects that involve the entirety of the ZTL dimer structure. The results provide a unique insight into the ZTL allosteric mechanism that has impacts on the broader LOV community, as well as providing atomistic details on how different allosteric variants can perturb signaling through regulation of ZTL dynamics. We briefly discuss the consequences of these studies below. Proximal signaling event and a Gln-less mechanism First, the computational tools employed herein verify proximal allosteric events identified in structural studies of ZTL allosteric variants. Namely, in the dark-state, Gln154 samples heterogeneous conformations that permit a dynamic dark-state landscape. Adduct formation triggers organization of Gln154 in a predominantly buried-II conformation that is coupled to the concerted movement of Phe66 and Phe156 stabilized by π-bond formation, and a new H-bond between Gln154 and Gly46. [50] The net result is ordering of the N/C-termini and formation of a light-dependent salt-bridge between Arg125-Glu158. We note that the observed allosteric process is directly analogous to recent computational studies of YtvA, [24] where they observed coupling between Gln conformational dynamics and light-dependent movement of a Phe residue equivalent to Phe66 (Phe46 in YtvA). These observations raise the possibility of a Gln-independent signaling mechanism that relies on Phe movement in response to changes in flavin electronics. The experimental studies conducted by Pudasaini et al [23] and Pérez et al [51] support this hypothesis by showing the ability of ZTL and YtvA to signal in absence of the Gln154. To verify such a mechanism, we examined MD simulations of Q154L variants to determine whether Phe movement independent of Gln154 can mimic the ZTL allosteric response. Indeed, a preliminary analysis showed that in Q154L variants the adduct formation was coupled to the movement of Phe156 towards flavin O4 position in a manner analogous to the photoactivation mechanism of WT ZTL (S18 Fig). The reason for this movement can be identified by the interaction between Phe66 and FMN, observed via distance analysis in S19 Fig, which stabilizes Phe156. Thus, the Phe156 movement alone is sufficient for signaling and can persist in the absence of Gln154. We propose that a similar electronics-based signaling mechanism can occur in other LOV proteins containing Gly residues at positions equivalent to G46. Importantly, Gln➔Ala variants retain some light-regulated activity in YtvA, consistent with an electronics-based mechanism. [51] Allosteric response Structural studies of WT ZTL and allosteric variants present a snapshot of ZTL in various functional states. However, they cannot provide insight into how local conformational changes propagate to a global rearrangement in a dimer interface. Further, traditional MD approaches alone are insufficient to capture a long time-scale reorganization of protein interactions (e.g. dimer interface). Here, we coupled traditional MD simulations in diverse functional states, and transient intermediates with MSM and ML classification models to capture how atomistic proximal events dictate dynamics of signaling motifs gating a conformational response. These approaches provide detailed information on the ZTL signaling mechanism and highlight structural motifs previously of unknown function. MSM and ML classification models indicate that proximal structural changes within the CGF (N-terminal) and QFF (C-terminal) motifs dictate ordering of N-and C-termini. Central to N/C-termini ordering is formation of a salt-bridge between Arg125-Glu158 connecting the C-terminus to the helical LOV domain surface. ML classification models indicate that a central component of ZTL signaling and conversion between anti-parallel and parallel configurations is ordering of the EF-loop. Previous low-resolution studies of the ZTL family member flavinbinding, KELCH REPEAT, F-BOX 1 (FKF1), had identified the EF-loop as involved in lightdependent dimer reorganization via an unknown mechanism. Here, we demonstrate for the first time direct coupling of adduct formation to allosteric events within the EF-loop. Notably, the EF-loops are highly conserved in ZTL and FKF1 proteins, but divergent between the two family members. Thus, it is likely that the EF-loop may be important in differentiating signaling members within this family. Further studies of the EF-loop and EF-loop variants can clarify how these underlying characterized structural motifs dictate LOV signal transduction and regulation of protein-protein interactions. Notably, the EF-loop is unique to members of the ZTL family, and fungal circadian clock photoreceptors VVD and White-Collar 1. Computational studies of both families now highlight coupling between N-terminal signaling regions and EF-loop dynamics, however with reversed topology. Specifically in ZTL, light drives ordering of the N/C-termini and resultant restriction on EF-loop dynamics. In contrast, although VVD was found to follow the same two steps, they occur in the reverse order. [19] The variation in the order in signaling events is consistent with the unique signaling mechanism of ZTL, where light represses the activity of the C-terminal E3-ligase activity in contrast to light-regulated increases in activity of most LOV proteins. Overall, our combined MSM/ML platform allows unique insight into how the LOV signaling mechanism can follow conserved pathways but modulate both the direction and magnitude of signaling responses depending on residue substitutions at key allosteric sites. Allosteric variants and alteration to dynamics-based signaling Structural approaches to identify allosteric mechanisms often rely on identifying allosteric variants that can trap the signaling protein in various presumed signaling states. In ZTL, proposed mechanisms were dependent on an allosteric variant (V48I:G80R) trapped a dark-state like conformation and an allosteric variant (G46S:G80R) trapped in a putative light-state like conformation. Although these structural approaches provide keen insights into both proximal and global conformational changes mediating signal transduction, their dependence on staticstructures masks dynamics-based impacts on signal transduction, and may reflect artifacts due to an altered signaling mechanism caused by the mutations. In ZTL, this led to an unexplained aspect of the G46S variant, that if G46S:G80R already adopts the light-state parallel dimer, how does light enhance protein-protein interactions with its target GIGANTEA? Here, a combined MD/MSM/ML platform provides direct insight into how allosteric variants alter protein dynamics and perturb a dynamics-based allosteric mechanism. We demonstrate how two variants have different effects on signal transduction. In the case of V48I, perturbation of the allosteric mechanism results directly from impeding Gln154 conformational dynamics necessary for sampling N/C-termini conformational states which are necessary for downstream allosteric responses. In contrast, G46S variants directly alter the entirety of the allosteric mechanism leading to sampling of only the light-state conformational landscape. In this altered mechanism, the steric bulk of Ser46 alters the interaction between Gln154 and Phe156, which perturbs both C-and N-termini dynamics. The altered interactions permit sampling of a unique (extended) Gln154 conformation that enhances sampling of the light-state landscape and leads to a light-dependent enhancement in GI interactions. In both cases, the variants selected are modest mutations widely tolerated in the LOV protein family, but uniquely evolutionarily selected for in ZTL. The ability of two residues in close proximity to direct unique alterations in signal transduction highlights the importance of combined structural and computational approaches in designing or studying new members of domain families. Molecular dynamics simulations The initial structures of the dark and light states for the ZTL dimer complex were taken from the Protein Data Bank (PDB): the wild type (WT) dark state in anti-parallel conformation (PDB ID: 5SVG), the light and dark anti-parallel conformations of V48I:G80R (PDB ID: 5SVV and 5SVW, respectively), and the G46S:G80R mutant in light parallel conformations (PDB ID: 6WLP). The light state WT anti-parallel and transient dark anti-parallel structures were constructed based on the light V48I:G80R mutant crystal structure. The WT Parallel structures were constructed based on the G46S:G80R mutant crystal structure. From the native dark and light crystal structures, two transient structures were generated. The transient dark structure was generated by breaking the covalent bond between the Cys82 and the FMN, in the native light structures of ZTL. The transient light structure was generated by forming the Cys82-FMN bond and protonating FMN N5 in the dark state structures of ZTL. The structures subjected to simulations in this study are listed in Table 1. To maintain the same number of residues for all structures, the termini of the monomers in native dark and light states were modeled to be the same length. The total number of residues per monomer is 130. All structures subjected to simulations in this study contain FMN as cofactor. The parameters used for the cofactor were taken from a previous study. [22] Hydrogen atoms were added to the crystal structures. The protonation states for the histidines were assigned using the ProteinPrepare tool of PlayMolecule. [52] The structures were then solvated using TIP3P water molecules and neutralized by adding chloride anions and sodium cations. The structures were minimized first with the steep descent method for 200 steps and the adopted basis Newton-Raphson minimization for 1000 steps afterwards. An initial 24 ps molecular dynamics was carried to increase the temperature from 0K to 300K. For each structure, three 10 ns isothermal-isobaric ensemble (NPT) equilibration dynamics started with random velocities were carried out. The final coordinates and velocities were used to start a production phase of 600 ns simulations of constant-volume constant-temperature (NVT) ensemble. In the production simulations, the first 100 ns were considered as equilibration and therefore excluded from the analysis. A total of 15 μs of MD trajectories have been generated. The simulations were carried using OpenMM 7.3 on GPU CUDA accelerated. [53] The NPT simulations were performed using a MonteCarloBarostat implemented in OpenMM 7.3. The NVT simulations were performed using the Langevin integrator. For the integrator, a friction coefficient of 1 ps -1 was implemented. [53] In all simulations, the SHAKE constraint for hydrogen covalent bonds was applied. [54] A step size of 2 fs was used. A triclinic box was used in the simulations. Period boundary conditions were applied. The particle mesh Ewald method was used to calculate the electrostatic interactions. [55] Trajectories analysis The convergence and the stability of the trajectories were assessed through the analysis of the Root Mean Squared Deviation (RMSD) of the Cartesian coordinates of all atoms between each frame from the simulation superimposed onto the first frame of the simulation as the reference structure. The RMSD for a given simulation is defined as: RMSD ¼ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi where N is the number of atoms, r i and r i 0 are the Cartesian coordinate vectors for atom i in a simulated conformation and reference structure, respectively. U is the best-fit alignment transformation matrix between a given structure and its reference structure. The Root Mean Squared Fluctuation (RMSF) of atom i is calculated as its averaged fluctuation. RMSF i ¼ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffiffi 1 T where T is the number of total frames, r j i is the coordinate of atom i in frame j, and � r i is the average position of atom i in the given trajectory. The RMSD and RMSF analyses, together with the distance analyses presented in this study, were performed using the MDtraj analysis package implemented in python. [56] The hydrogen bonds analysis presented in this study was carried out using a python script to detect hydrogen bonds according to the Baker and Hubbard criteria. [57] The visualization of the protein structures in the different analysis was done using 3D Protein Imaging [58] and VMD 1.9.3. [59] Time-structure independent component analysis Time-structure independent component analysis (tICA) was employed as the dimensionality reduction method to analyze the simulations. The tICA method maximizes the auto-correlation function, which results in finding the slowest degrees of freedom and therefore in preserving the kinetic information present in the trajectories. [60][61][62][63] Given a time-series of molecular coordinates provided by the MD trajectories x(t) = t (x 1 (t),. . .,x n (t)), tICA aims to reduce the dimensionality of the trajectories and to identify hidden key structural changes by decomposing the generalized eigenvalue problem � CF = CFK. Where K = diag(k 1 ,. . .,k n ) and F = (f 1 ,. . .,f n ) are the eigenvalue and eigenvector matrices, respectively. C and � C are the covariance matrix and the time-lagged covariance matrix of the coordinate vector, respectively. C ¼ hðxðtÞ À hxðtÞiÞ t ðxðtÞ À hxðtÞiÞi ð3Þ In order to obtain a symmetric time-lagged covariance matrix, 1 Þ is calculated. The latter step assumes the time reversibility of the process, which is satisfied in MD trajectories. Lastly, the projected vectors of the MD are: The tICA method has been successfully applied for the kinetic analysis of MD simulations. [60][61][62][63][64] In this study, the first two components from tICA were selected for visualization, interpretation and further analysis. The identification of the tICA component was based on all the trajectories considered in this study. In computing the tICA components, the trajectories were first concatenated and 100 ps was used as lag time for maximal inclusion of atomistic correlated motions (S20 Fig). The featurization of the protein using all the Cα pairwise distances and dimensionality reduction were performed using the MSMBuilder package. [65] Markov state model Markov state models (MSMs) [29][30][31][32][66][67][68] have become increasingly useful network models with the continuously developing open source software infrastructure [65,[69][70][71] for describing the transitions among functional states during allosteric events. [72][73][74] Combined with MD simulations, MSM approaches can provide connectivity maps of states on the free energy landscape, estimate the effect of allosteric perturbations in the conformational equilibrium, and rigorously describe kinetics of allosteric transitions. Recent advances in the field have highlighted how MSM tools can help to recognize structural and dynamic patterns of conformational ensembles, identify functional allosteric states hidden in the conformational ensembles and reconstruct allosteric mechanisms. [74] Markov models have been employed for understanding the reaction mechanisms, thermodynamics and free-energy landscape population shifts, the hierarchy of timescales and the structural basis of allosteric events. [19,[29][30][31][32][33][34] MSM is used to track the conditional transition probabilities among non-overlapping states. The collection of the transition probabilities among n states is the transition matrix , where c ij is the count of the number of times the trajectories transition from a state i to a state j within a certain time interval Δt, called lag time τ. To build the MSM, the trajectories were clustered based on geometrical similarity into 150 clusters (S8 Fig) using MiniBatchKmeans implemented in MSMbuilder. The clustered trajectories were then fed to MSMBuilder 3.8.0 to construct kinetically relevant Markov states. Using the transition matrix, the probability to find the system in any state after certain time τ can be calculated using the equation p(t+τ) = p(t)T(τ). The first two components of tICA were used as collective variables. MSMBuilder python package was used to build the MSM. [65] The default hyper-parameters provided by MSMBuilder were used for the analysis. The ergodic cutoff was turned on and the Maximum Likelihood method was used to achieve the reversibility of the transition matrix. An appropriate lag time for transitions among microstates was estimated based on the convergence of the implied relaxation timescale. A relaxation time scale can be interpreted as a time length needed for the system to reach a particular steady state. [75] The implied relaxation timescales obtained at various lag times show the convergence roughly for lag times longer than 30 ns (S8 Fig), which is chosen to build the final MSM. A total of 11 macrostates were created using the Perron-cluster cluster analysis (PCCA) implemented in the MSMbuilder package. [65] This method is based on the assumption that states belonging to the same free energy basin will interconvert easily, providing a higher transition probability for these states. Therefore, the separation of the states can be extracted via spectral decomposition of the transition probability matrix. [76] Machine learning The supervised machine learning method one-vs-one (OvO) random forest [77] implemented in the Scikit-learn python package [78] was used in this study. OvO Random Forest uses parameters F = (j,t) composed of the data features j and a threshold t to divide the data in two parts based on the threshold. with x being the training data and y being the training label. The protein features utilized for this analysis were all the Cα pairwise distances. The Gini impurity criterion was used to assess the quality of the model. The Gini impurity score represents the likelihood of an incorrect classification of a new random variable according to the existing distributions of the labels k: Decision tree methods suffer from possible bias towards certain set of features given a certain data batch. [79] The random forest approach overcomes this problem and ensures more statistical solidity by taking a collection of decision trees, which are composed of different data batches, and uses a majority voting system amongst the classification output of all the single trees. [80] OvO random forest model is trained not on all the classes but for each pair of classes. [81] This provides the possibility to extract not only the features that are responsible for the distinguishability of the different classes (metastable states) and their importance, but also to compare each pair of metastable states as well. The number of estimators for an OvO Random Forest model is: A total of 55 estimators were built. The validation of the Random Forest hyper parameters was performed using the GridSearch method implemented in Scikit-learn. The values which provided the best accuracy for our model are 10 trees of 4 layers depth. Pearson correlation analysis The Pearson correlation (PC) is a measure of linear correlation between two variables. [82] The first step in building the PC is calculating the covariance between two variables: The covariance is then divided by the square root of the product of the variance of each variable. r x; y ð Þ ¼ sðx; yÞ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi VarðxÞVarðyÞ p ð11Þ PC is dimensionless and can assume values in the range of [-1, +1], where the two extremes stand for total anti-correlation and total correlation, respectively.	2021-07-28T06:17:55.251Z	2021-07-01T00:00:00.000	{ "year": 2021, "sha1": "c29161d85bb9dd376d15c8f7746441735c9b5239", "oa_license": "CCBY", "oa_url": "https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1009168&type=printable", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "429a887b2aa87fe438f1f83731a0360ad2420cd5", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
14164324	pes2o/s2orc	v3-fos-license	Decoupling of the Right-handed Neutrino Contribution to the Higgs Mass in Supersymmetric Models Recently, it has been argued that in the supersymmetric extension of the seesaw-extended Standard Model, heavy right-handed neutrinos and sneutrinos may give corrections as large as a few GeV to the mass of the lightest neutral CP-even Higgs boson, even if the soft supersymmetry-breaking parameters are of order the electroweak scale. The presence of such large corrections would render precise Higgs masses incalculable from measurable low-energy parameters. We show that this is not the case: decoupling is preserved in the appropriate sense and right-handed (s)neutrinos, if they exist, have negligible impact on the physical Higgs masses. Introduction The discovery of a new boson near 126 GeV [1] that resembles the Higgs boson of the Standard Model has stimulated considerable theoretical interpretation. In supersymmetric models, the observed mass is particularly interesting. Whereas 126 GeV is compatible with expectations for the mass (m h ) of the lightest neutral CP-even Higgs boson of the Minimal Supersymmetric Standard Model (MSSM), large quantum corrections are indicated in order to raise m h to a value 40% above m Z [2]. In the next-to-minimal model (NMSSM), an additional tree-level contribution may also boost the value of m h , but radiative effects are still necessary unless the tri-linear coupling of the singlet and doublet Higgs fields in the superpotential is large [3]. Thus the measured Higgs mass provides an important clue to the parameters of the supersymmetric model. The program of precision calculations of the lightest Higgs boson mass in the MSSM began with one-loop results, given in [4], followed by two-loop contributions given in [5]. Partial three-loop results are now available [6]. In cases with a large hierarchy between the weak scale and the scale of the stop squarks, resummation has been used to obtain precise results, now at the level of three-loop β-functions in some cases [7]. Residual theoretical uncertainty estimates vary depending on the type of calculation performed, but in the fixed-order case are perhaps of the order 1 GeV for light spectra below a TeV, and 2-3 GeV for heavier spectra [6]. The utility of the computations described above rely on decoupling -very heavy states that do not receive their masses from electroweak symmetry breaking are expected to give negligible contribution to the lightest Higgs mass. Only a limited set of model parameters, which are in principle accessible at future collider experiments, are thought to be required for an accurate calculation of m h . On the other hand, if an inaccessible heavy sector could provide a significant contribution to m h , then only the size of this contribution could be constrained by comparing the measured m h to the calculation in terms of observable parameters. This would clearly be a much weaker position. Recently, it has been suggested in Ref. [8] that in the seesaw-extended MSSM [9,10], a righthanded neutrino and sneutrino provides an example of such a non-decoupling heavy sector, potentially shifting the MSSM prediction for m h by as much as a few GeV at one-loop order, even if the soft supersymmetry (SUSY) breaking parameters remain at the TeV scale. It was further argued that the large terms appear at order p 2 in the relevant two-point functions, which are invisible to effective potential estimates that are based on calculations performed at zero external momenta. 1 In light of its importance for the interpretation of the observed Higgs boson with m h 126 GeV, we have performed a reanalysis of the right-handed neutrino and sneutrino contributions to m h in the seesaw-extended MSSM. We find that the corrections to m h due to physics at the seesaw scale are always minuscule, of the order of a billionth of an eV. This decoupling behavior is manifest in renormalization schemes in which the tan β counterterm is completely insensitive to phenomena at scales well above the SUSY-breaking scale. One class of decoupling schemes includes physical schemes, where the tan β counterterm is controlled, for example, by the radiative corrections to the mass of the heavy Higgs boson, or by corrections to the decays of the heavy Higgs bosons to down-type fermions. Another class of decoupling schemes subtracts non-decoupling terms by hand, mocking up the behavior of minimal subtraction schemes where heavy particles are fully integrated out at their thresholds and are absent from the low-energy theory. If a non-decoupling renormalization scheme is employed in the definition of tan β, then the decoupling of high-scale physics phenomena in the radiatively-corrected Higgs mass is recovered once tan β is directly related to a low-energy observable. That is, tan β should be regarded as an intermediary quantity, which one is free to define in any scheme. Independently of how one defines tan β, the MSSM Higgs mass ultimately depends solely on parameters that can be fixed by experimental measurements at energy scales of order the SUSY-breaking scale and below. Contributions to the Higgs mass from energy scales significantly above the SUSY-breaking scale must be negligible. This paper is organized as follows. In Section 2 we review the computation of the physical masses of the neutral CP-even Higgs bosons in the MSSM at one-loop order. We provide compact formulae and discuss the role of tan β renormalization in the results. In Section 3 we calculate the leading contributions to the lightest Higgs boson mass from the left and right handed neutrino/sneutrino sectors. We reduce the full diagrammatic result of Ref. [8] to simple, approximate analytic formulae in two different renormalization schemes, and find that in both cases the right-handed neutrino sector exhibits appropriate decoupling. We provide an interpretation of our approximate formulae in the more natural setting of effective field theory. Finally, we study the full one-loop results numerically, finding again that contributions from the right-handed neutrino sector are negligible. Our conclusions are presented in Section 4. Explicit expressions for self-energy functions, tadpoles and the tan β counterterm, which can provide potential non-decoupling contributions in the computation of the Higgs mass, are exhibited in Appendix A. Using these approximate forms, one can check that the non-decoupling terms cancel exactly in the expressions for the one-loop radiatively-corrected Higgs mass when a suitable definition of the tan β counterterm is employed. Physical Higgs Masses at One Loop in the MSSM We begin our discussion with a review of the one-loop physical Higgs masses in the MSSM with the minimally required two-Higgs doublet Higgs sector. The neutral field content is Here v u [v d ] are the vacuum expectation values (vevs) of the neutral Higgs fields that couple exclusively to the up-type [down-type] quark and lepton fields. The MSSM Higgs scalar potential is given by µ is the supersymmetric Higgs mass parameter, and m 2 u , m 2 d , and b are soft SUSY-breaking squared-mass parameters. The linear terms in the potential are given by: The quadratic terms yield 2 × 2 scalar and pseudoscalar squared-mass matrices [in the (φ d , φ u ) basis], All parameters appearing in the above formulae should be interpreted as bare parameters. It is convenient to require that v u,d are stationary points of the full one-loop effective potential, which is achieved via the tadpole cancellation conditions, The functions A u,d are the one-loop tadpole diagrams at zero external momentum, and the T u,d are functions of the bare parameters given in Eqs. (2.3) and (2.4). Using Eq. (2.7), the pseudoscalar mass matrix simplifies to Diagonalizing this matrix and expanding to leading order in A u,d , the bare masses for the pseudoscalar A and the Goldstone boson G are found: Inserting these results into M 2 e , we obtain where m 2 Z ≡ 1 2 G 2 v 2 , s β ≡ sin β, and c β ≡ cos β. The squared-mass matrix M 2 e can be diagonalized to obtain the bare masses m 2 h,H for the light neutral CP-even Higgs boson h and the heavy neutral CP-even Higgs boson H. At this stage, it is convenient to replace the bare masses by physical masses: h,Z,A,H = m 2 hP,ZP,AP,HP − Σ hh,ZZ,AA,HH (m 2 hP,ZP,AP,HP ) , (2.13) where the subscript P indicates the corresponding physical parameter. The Σ functions are the real parts of the corresponding self-energy functions 2 through which parameters from other sectors of the theory affect the Higgs masses. At one-loop order, the arguments of Σ hh,HH can be consistently replaced with the corresponding tree-level expressions for the physical masses, 14) The replacements of Eq. At one-loop order the renormalization of the vevs affects the Higgs masses only through the parameter and (2.19) where m ht,Ht are the tree-level masses of the neutral CP-even Higgs bosons [cf. Eq. (2.14)], m A and m Z are the physical masses 3 (i.e., input parameters taken from experimental measurements), the angle α is the tree-level mixing angle obtained in the diagonalization of M 2 e , and are the tadpoles with respect to the neutral CP-even Higgs mass basis. In obtaining the formulae in Eqs. (2.18) and ( , (2.20) as well as the relation between the tadpoles and the Goldstone self-energy imposed by the requirement that the one-loop Goldstone boson mass vanishes, As a check of our calculation, we note that in the limit of β = 1 2 π and m A > m Z we have m ht = m Z , m Ht = m A , and sin(β − α) = 1 at tree-level. In this case, Eqs. (2.18) and (2.19) reduce to which reproduces the result for m h obtained in Ref. [4]. From Eqs. (2.18) and (2.19), we see that the only counterterm appearing explicitly in the Higgs masses is δ tan β. If only a prediction for m h is desired, then δ tan β can be eliminated in favor of m H , and all instances of the renormalized tan β parameter appearing in the self-energies may be consistently replaced at one-loop order by solving the tree level formula Eq. (2.14) for tan β as a function of m H . The end result, coincides with a sum rule derived first in Ref. [14]. In the MSSM, the prediction for m h and m H depends on tan β and other MSSM mass parameters (such as m A and the top squark mass and mixing parameters). In particular, since tan β appears in the expressions for m ht and m Ht [cf. Eq. (2.14)], one must define δ tan β by specifying a subtraction scheme. In principle any scheme to define the parameter tan β is allowed. In practice, it is preferable to employ a scheme that satisfies decoupling, in which case tan β can be determined solely from physical measurements that can be carried out in collider experiments. In contrast, if a non-decoupling scheme is used, then the definition of tan β depends on unknown contributions from inaccessible heavy sectors, in which case the value of tan β (which is needed to predict m h and m H ) cannot be determined from low-energy experimental measurements. Of course, in the context of a specific model of high scale physics, one can employ a non-decoupling scheme to define tan β and then compute the relation of tan β so defined to some specific low-energy observable. In this case, one can formally eliminate tan β and re-express the MSSM prediction for m h and m H in terms of the corresponding low-energy observable. This would then provide a prediction for m h and m H in terms of parameters that can be determined solely from low-energy measurements. Following such a procedure, one finds that the predicted values for m h and m H are completely insensitive to high-scale physics, as expected from the decoupling properties of quantum field theory (e.g., see Ref. [13]). By employing a definition of tan β that respects decoupling, the insensitivity of the predicted values for m h and m H to high scale physics is manifest. Suppose that there are no schemes in which tan β can be determined from a low-scale measurement. As a simple example, consider the case of high-scale SUSY in the decoupling limit, where all the superpartner masses and m A are taken very large, of order m SUSY m Z . In this case decoupling schemes for tan β are not particularly favored over non-decoupling schemes. On the other hand, the observed Higgs mass is no longer a testable prediction, but rather a scheme-dependent constraint on the two unmeasurable parameters m SUSY and tan β. Scheme-dependence is not very important in this case as it can simply be absorbed in an unobservable shift of tan β. Furthermore, it does not affect the upper bound on m h for fixed m SUSY , which is obtained in the large tan β limit where scheme-dependent terms vanish. For the rest of this work, we will focus on the case in which the MSSM Higgs mass prediction is testable at colliders. The standard DR scheme [15] will not automatically yield decoupling. However, it can be modified slightly (mDR, in the notation of Ref. [8]) to remove large logarithms by hand. This subtraction reproduces the result one would obtain at leading-log order with effective field theory, in which heavy sectors are integrated out by hand at their thresholds. Hence, at leading-log order the mDR scheme respects decoupling. However, beyond leading-log, one should also remove non-decoupling non-logarithmic finite terms that are still present in the mDR scheme. This can be achieved in an extension of the mDR scheme in which all contributions from the heavy sector are subtracted. A scheme that possesses similar properties, denoted by "DEC" (for decoupling) below, fixes the Higgs wave function counterterms as follows, 4 In this scheme, the tan β counterterm is given by Eq. (2.17), Indeed, the DEC scheme manifestly removes large logarithms and finite terms from heavy sectors (as we exhibit explicitly in Section 3.1). This subtraction scheme also removes additional contributions that depend on the low-energy sectors (without affecting the decoupling behavior of the scheme). In fact, this is reminiscent of the on-shell scheme (the definition of which does not involve the limit α → 0) which was observed in Ref. [8] to respect decoupling, but was discarded in favor of the mDR scheme, as the latter was deemed to be more numerically stable. We emphasize that even with a scheme (such as the DEC scheme) that is not directly related to any particular physical measurement, decoupling is preserved if the effects of the heavy sector that do not vanish in the large mass limit are fully removed by hand. In particular this is how effective field theory analysis should be performed in mass-independent schemes [16]. Another possibility is to demand that some physical (measurable) quantity is given at one-loop order by its tree-level formula. Two such quantities are the mass m H and the decay rate Γ(A → τ τ ). In the former case [denoted as the "HiggsMass" (HM) scheme], the tan β counterterm is obtained by setting m 2 H = m 2 Ht in Eq. (2.19), which defines tan β in terms of the low-energy physical parameters m Z , m H and m A , so that all one-loop pieces cancel: A detailed and complementary discussion of tan β renormalization appears in Ref. [17]. In this reference, the authors do not emphasize decoupling properties, but exhibit other flaws among all available schemes. For example, DR is gauge-dependent at one-loop, the HM scheme can lead to large perturbative corrections and numerical instability, and using Γ(A → τ τ ) is both technically complicated and introduces flavor dependence into tan β. For our purposes of exhibiting decoupling in the next section, we will use the DEC and HM schemes as examples. Regardless of the scheme used to define δ tan β, measuring Γ(A → τ τ ) is a good way to experimentally determine the numerical value of renormalized parameter tan β in the given scheme. Once tan β, m A , and the soft parameters are fixed (either by hand or from experimental determinations), m h and m H become predictions of the theory. Right-Handed Sneutrino Contributions to m h Right-handed neutrinos and sneutrinos obtain supersymmetric masses and couple to the Higgs sector through the following superpotential interactions [9,10]: 1) where N and R represent the right-handed neutrino and lepton multiplets, respectively, and m M is the Majorana mass. There are also new soft SUSY-breaking couplings and masses given by the potential In general all masses and couplings are 3 × 3 matrices in flavor space, but for simplicity we consider only a single flavor. The resulting neutrino mass matrix is given by where m D ≡ y ν v u . The CP-even/odd (+/−) sneutrino mass matrices are given by [9]: 4) where m 2 L is the usual soft-breaking mass for the left-handed sneutrinos present in the MSSM. In the analysis presented in this paper, we consider only a single flavor of right-handed neutrinos and sneutrinos as described above. Nevertheless, our conclusions should not be affected by the presence of additional generations of right-handed neutrinos and sneutrinos. Approximate Diagrammatic Result We expect that the right-handed neutrino and sneutrino contributions to the physical Higgs masses should decouple as the Majorana mass scale becomes much larger than the soft supersymmetry breaking scales, if all other parameters are held fixed. This expectation is based on the fact that the Majorana mass term m M that appears in the superpotential [cf. eq. (3.1)] is a supersymmetry-preserving parameter. Indeed it is well known that the corrections to the tree-level Higgs mass relations in the MSSM are due entirely to SUSY-breaking effects. In contrast, we do not expect decoupling if the SUSY-breaking parameters associated with the right-handed sneutrino sector are taken very large. In the calculations presented in this section, we shall initially assume that all SUSY-breaking masses are no larger than O(1 TeV). The consequences of large SUSY-breaking in the right-handed sector will be briefly considered in Section 3.4. The relevant one-loop tadpoles and self-energy functions are given in the appendix of Ref. [8]. We have independently computed Σ hh,ZZ and A h in the cot β → 0 limit and found agreement except for the minus signs in front of the m 2 Z terms in the last and third-to-last lines of Eq. (81) of Ref. [8]. Inserting the formulae for the one-loop tadpoles and self-energy functions into Eqs. (2.18) and (2.19), we obtain the full results for m 2 h,H . To avoid a proliferation of scales tangential to the question of decoupling, we turn off A ν − µ cot β and the B ν parameter, and fix a common scale m S , where mL = mR ≡ m S . 5 We expand to first order in m 2 /m 2 M , where m ∈ {m Z , m S , m D }, and to leading order in powers of m Z , which is the smallest mass scale when the superpartner masses, the CP-odd Higgs mass m A , and the Dirac mass are large. Note that keeping only the leading order in m Z is equivalent to taking α β − π/2 (since the vev v aligns with the light state h in this limit). At leading-logarithmic order, we find that the lightest Higgs mass squared is shifted relative to its tree level value in the two renormalization schemes by an amount The first terms on the right-hand side of Eq. (1), it follows that m D cannot be larger than the electroweak scale, in which case m ν is at most of order 1 eV for a suitably chosen right-handed neutrino mass scale. Hence, the magnitude of the corrections to m h due to the right-handed neutrino sector are always minuscule. For the calculation of ∆m 2 h in the HM scheme, we avoided the direct computation of δ tan β by taking advantage of the sum rule, substituting everywhere the tree level expression for m 2 H . Therefore, as a check of Eq. (3.5), we can compute the relation between tan β in the two schemes and see if it is consistent with the difference in the two computations of ∆m 2 h . The relation between the renormalized tan β parameters is determined by the counterterms, tan β HM = tan β DEC + δ tan β HM − δ tan β DEC , (3.6) where δ tan β DEC is given by Eq. (2.25) and δ tan β HM is given by Eq. (2.26). Hence, the shift in the one-loop prediction for m 2 h incurred by changing schemes is given by inserting Eq. (3.6) into the tree level formula for m 2 h : We find, in the approximations used above for ∆m 2 h , δ tan β HM − δ tan β DEC tan β cos 2β It is straightforward to check that inserting Eq. where Q is the renormalization scale. As noted in Ref. [8], the partial decoupling-by-hand of the mDR scheme can be achieved in the DR scheme by taking Q 2 = m 2 M . However, a finite non-logarithmic term remains that also must be subtracted by hand if tan β is to be a genuine low-energy parameter that can be determined from experimental measurements far below the seesaw scale. Indeed, one could simply extend the mDR scheme by performing this extra subtraction. The end result is equivalent to the DEC scheme at leading order in our expansions. To make further contact with the results of Ref. [8], we first note that Eq. (2.18) can be rewritten as where Σ hh (p 2 ) is defined in Eq. (3.7a) of Ref. [8]. 6 If the two-loop contributions generated by products of self-energy functions are neglected in Eq. (3.2) of Ref. [8], then the pole in the matrix propagator corresponding to the light CP-even Higgs mass is given by where m 2 h appearing on the right-hand side above is the one-loop corrected Higgs mass. Note that the fact that the argument of Σ hh is m 2 h rather than m 2 ht means that partial two-loop information is being included in the expression for the one-loop corrected Higgs mass. In this case, Eq. (3.7a) of Ref. [8] implies that the loop-corrected Higgs mass given by Eq. (3.11) is equivalent to Eq. (2.18) with the following replacement, We now examine in more detail how decoupling occurs in the expression for the loop-corrected Higgs mass. It is convenient to define a momentum-dependent Higgs squared-mass, 3.14) where m 2 h (p 2 ) corresponds to the result of Eq. (2.18) after replacing Σ hh (m 2 ht ) with Σ hh (p 2 ). By choosing either p 2 = m 2 ht or p 2 = m 2 h , we recover either Eq. (3.10) or Eq. (3.11), respectively. The potential non-decoupling behavior lies in the O(m 2 D ) contributions to the loop-corrected Higgs mass. In Appendix A, we give the leading terms contributing at O(m 2 D ) in the individual self-energy functions, tadpoles, and the tan β counterterm. None of the individual terms that appear in the expression for the loop-corrected Higgs mass vanish in the large m M limit. However, given a decoupling scheme for δ tan β [and δZ hh , if Eq. (3.12) is used], then the non-decoupling terms cancel exactly in the Higgs mass prediction, leaving only m 2 M -suppressed terms at O(m 4 D ). It is instructive to evaluate the O(m 2 D ) contributions to m 2 h (p 2 ) in the DEC scheme. Using the results of Appendix A, we readily find that (3.15) where The end result is To complete the computation of m 2 h (p 2 ), we make use of Eqs. (2.24) and (3.13) Using Eq. (3.14), it follows that the O(m 2 D ) contributions to m 2 h (p 2 ) exactly cancel in the DEC scheme. This decoupling has already been demonstrated for the one-loop corrected Higgs mass defined by Eq. (2.18) in the DEC scheme [cf. Eq. (3.5)]. One can repeat the above calculation in the HM scheme, where m 2 h (p 2 ) is most easily obtained using Eq. (2.23), which yields Evaluating the self-energy functions using the results of Appendix A, we again recover the result of Eq. (3.15). For p 2 = m 2 ht , the O(m 2 D ) terms vanish exactly and the decoupling behavior is established, as previously demonstrated. In the case of p 2 = m 2 ht , we need a separate definition of the Higgs wave function counterterms. Here, the natural choice is an on-shell scheme, which fixes the residues of the corresponding pole masses to unity. In this scheme, the O(m 2 D ) contributions to δZ hh O(m 2 D ) are the same as those of the DEC scheme, since the O(m 2 D ) contributions to dΣ hh (p 2 )/dp 2 and dΣ HH (p 2 )/dp 2 are independent of p 2 . Thus, it again follows that the O(m 2 D ) contributions to m 2 h (p 2 ) exactly cancel in the HM scheme. (3.20) To obtain the corresponding DR expression for δZ hh O(m 2 D ) , we retain −1 − γ + log 4π in Eq. (3.18) and discard the remaining terms. Thus in the DR scheme, Eq. (3.14) yields In the mDR scheme of Ref. [8], one sets Q 2 = m 2 M to remove the logarithm, but the constant term remains and decoupling is not satisfied. The loop-corrected Higgs mass advocated in Ref. [8] corresponds to setting p 2 = m 2 h in m 2 h (p 2 ) [cf. Eq. (3.11)]. In this case, there are two separate contributions to the non-decoupling behavior, corresponding to the two terms obtained in Eq. (3.21). In the mDR scheme, the second term of Eq. (3.21) is negative and provides the dominant source of the Higgs mass shift at large tan β. Indeed, it is of the correct order of magnitude to explain the decrement in m h obtained in the numerical analysis of Ref. [8]. Thus, we have located the sources of the non-decoupling behavior found in Ref. [8]. However, we note that even in a consistent one-loop truncation where p 2 = m 2 ht is taken to evaluate the loopcorrected Higgs mass, there is still a residual non-decoupling behavior in the mDR scheme, which enters via the tan β counterterm (which fixes the definition of tan β). In contrast, by employing a decoupling scheme to fix the tan β counterterm (and the Higgs wave function counterterms if separately needed), one is guaranteed a loop-corrected Higgs mass that is completely insensitive to the physics at the right-handed neutrino scale (assuming this scale lies significantly above the SUSY-breaking scale). Effective Field Theory Estimates of the Higgs Mass Shift In Ref. [8] it was argued that large corrections to m h could be traced to terms proportional to the external momenta in the self-energy functions. Such terms would not appear in the usual effective potential calculation. However, we have found that in a consistent one-loop truncation, such large corrections do not appear in the full expression for the physical Higgs mass when expressed in terms of parameters that can be measured directly in the low-energy effective theory. Therefore, it should be possible to derive the parametric properties of the leading terms presented in Section 3.1 directly from corrections to the Higgs quartic coupling in the effective potential, as computed in effective field theory (EFT)-the natural framework for dealing with large mass hierarchies. For simplicity, we will work primarily in the small-m Z limit, where the vev v aligns with the light state h such that α → β − π/2. The m 2 Z term we found in ∆m 2 h is just the usual contribution at low scales from the D-term coupling \|H u \| 2 \|L 2 , and is insensitive to the m M threshold. What about the subleading term? Imagine that we integrate out the right-handed neutrino and sneutrino at the right-handed neutrino mass threshold. Above this scale, the running of λ (the coefficient of the quartic self-coupling 1 8 h 4 in the effective Lagrangian) is supersymmetric, but the TeV-scale soft mass splits the scalar and fermion states, leading to a logarithmic correction to λ from the right-handed sneutrino bubble diagram: This term is certainly present in the corrections, but it is m M -suppressed and has no log enhancement, so it is not the source of the second terms in Eq. (3.5). In addition to direct contributions to λ, we also generate an approximately supersymmetric higher-dimensional coupling, Figure 1: Diagrams contributing to the running of the Higgs quartic below the right-handed neutrino mass scale. This coupling affects the running of λ when supersymmetry is broken via the diagrams in Fig. 1. The dominant contribution comes from the sneutrino diagram, (3.24) Running the quartic coupling down from m M to m S and recalling that v = √ 2m W /g, we obtain at leading logarithmic order, 3.25) matching the terms in Eq. (3.5) in the DEC scheme. To understand why we obtained the DEC scheme result instead of the HM scheme result, and how the latter can be reproduced, we have to consider the definition of tan β in the effective theory. Up to threshold corrections that are subleading (not log-enhanced), tan β EFT = tan β full at the matching scale Q = m M . Therefore, the tree-level boundary condition for the Higgs self-coupling λ takes the usual form, Below m M , the dimension-5 operator contributes to the running of tan β in a scheme-dependent way. It is straightforward to check that the beta-function for tan β does not contain terms proportional to m 2 S /m 2 M in the DEC scheme or any minimal subtraction scheme, where the field-strength renormalization counterterms are set by derivatives of self-energies with respect to p 2 . The relevant diagrams are obtained by setting two external legs to v u in Fig. 1, which makes it clear that the sneutrino loop is independent of p 2 . Therefore, in the DEC scheme, the corrections to m 2 h from the running of tan β are higher-order in the m Z expansion, and are not required to reproduce Eq. (3.5). In contrast, the tan β counterterm in the HM scheme is controlled by the self-energies themselves instead of their p 2 derivatives. At leading order in the m Z expansion, Eq. (2.26) with α = β − π/2 yields: Therefore, the sneutrino contributions to Σ HH (m 2 H ) and Σ AA (m 2 A ) can provide m 2 S /m 2 M terms in the running of tan β. Explicitly, (3.28) which at leading-log yields, Adding Eq. (3.29) to Eq. (3.25), we recover the leading HM scheme expression given by the full theory in Eq. (3.5). A more complete effective field theory analysis of the threshold corrections from the right-handed neutrino/sneutrino sector is beyond the scope of this paper. However, our full-theory calculation makes clear how decoupling will manifest at the thresholds. Loop diagrams involving right-handed neutrinos or sneutrinos will indeed provide non-decoupling finite contributions to the low-energy effective Higgs self-coupling λ during matching, but these contributions will be absorbed by finite and unobservable shifts in tan β. Numerical Results The full one-loop analytic formulae for the Higgs mass shifts in the decoupling schemes are too complicated to reproduce here. On the other hand, the approximations used above do not rule out the possibility of large corrections proportional to m 2 M or log m 2 M appearing at higher order in the m Z expansion or in non-logarithmic terms. To demonstrate that such terms are not present, we have numerically evaluated the full one-loop (s)neutrino contribution to m h as a function of \|m ν \| and m N , with the pure left-handed sneutrino contribution subtracted out. For definiteness, we define If we increase m M and m D so that the light physical neutrino mass m ν is fixed, we see that the corrections are roughly constant, also as expected. In either case the overall magnitude of the corrections is never larger than about 10 −10 eV, which is consistent with our estimate from Eq. (3.5). Large SUSY-Breaking in the Right-Handed Sector Consider the impact of choosing values for the SUSY-breaking parameters m 2 R and B ν that are large compared to the other SUSY-breaking parameters. If soft squared-mass parameter m 2 R becomes of order m 2 M , then the contribution to the Higgs quartic coupling from the running between m 2Ñ and m 2 N no longer decouples with large m M . The Higgs mass receives a correction of order 3.31) in complete analogy to the contribution from the top squarks. However, m 2 R also enters into the oneloop RGE for the Higgs mass parameter m 2 Hu , and therefore exacerbates the little hierarchy problem when m 2 R m 2 Z . For this reason it is preferable to keep m 2 R of the same order as other squark and slepton squared-mass parameters. When the soft mass parameter B ν dominates the SUSY-breaking parameters, it splits the CP -even and CP -odd right-handed sneutrinos according to mÑ ± ≈ m M ± B ν . It also alters the running of the Higgs quartic coupling at high energy scales and inhibits decoupling. Running between m 2Ñ + and m 2Ñ − yields a correction to the Higgs mass of order (3.32) The primary distinction from the case of large m 2 R is that B ν lowers the geometric mean of the righthanded sneutrino masses, making the logarithm negative and decreasing the Higgs mass. However, as in the case of m 2 R , there is a good reason to keep B ν m M . In particular, a large value of B ν generates a large contribution toν L -ν R mixing, which in turn generates a one-loop correction to the light neutrino masses that swamps the tree-level seesaw contribution if B ν 10 3 mν L [9]. In both the large m 2 R and large B ν scenarios, the contribution to m h from the left-handed sector diagrams of Fig. 1 are subdominant. The large right-handed neutrino-sneutrino mass splittings change the argument of the logarithm, but the contribution remains suppressed by the left-handed neutrinosneutrino mass splitting controlled by m 2 L . Conclusions A recent analysis [8] has argued that adding a right-handed neutrino and sneutrino to the MSSM could generate a sizable radiative contribution to the lightest Higgs boson mass in the case of a large righthanded neutrino mass scale, even if all soft SUSY-breaking parameters remain at the TeV scale. Such a non-decoupling effect would cast doubt on the notion that the Higgs mass can be reliably calculated in a weak-scale supersymmetric theory in terms of measurable TeV-scale parameters. In this paper we have reanalyzed the radiative corrections to the Higgs mass from the right-handed neutrino sector. In the analysis presented in this work, we began with a review of the computation of one-loop corrections to the physical masses of the neutral Higgs bosons of the MSSM, streamlining the derivation, providing compact general formulae for the spectrum, and reviewing the decoupling properties of various tan β renormalization schemes. In our consideration of the relevance of decoupling, we distinguished two cases. First, we commented briefly on the possibility that tan β cannot be independently measured in any scheme. For example, this could occur simply because all MSSM degrees of freedom are too heavy, in which case the decoupling properties of the scheme used to define tan β are irrelevant. However, the corresponding MSSM Higgs mass prediction cannot be tested, and the most that can be achieved is a scheme-dependent constraint on the superpartner mass scale and tan β. Much more relevant for phenomenology is the alternative case, where some MSSM particles with tan β-sensitive couplings can be accessed in collider experiments. In this latter case, one can predict the masses of the MSSM Higgs bosons in terms of quantities that are directly accessible to experimental measurements. These predicted masses are completely insensitive to physics at mass scales significantly larger than the scale of SUSY-breaking (such as the high-scale seesaw sector employed in a theory of neutrino masses). Consequently, it is especially convenient to define the parameter tan β using a renormalization scheme that respects decoupling, since the expressions for the MSSM Higgs masses (which depend explicitly on tan β) will then manifestly exhibit the expected decoupling behavior. Applying the general mass formulae to the right-handed neutrino sector, we derived expressions for the leading contributions in two decoupling schemes, and found that the magnitude of the corrections to the Higgs mass are utterly negligible. The expected decoupling behavior is observed if the righthanded neutrino mass scale is taken large while other input parameters are held fixed. The structure of the leading correction terms is easily recovered from effective field theory arguments. Finally, to go beyond the approximate formulae, we performed a numerical analysis including all contributing one-loop terms. We find that the corrections remain negligible and are well-reproduced by the leading terms. Since all the relevant couplings are weak, it is sufficient to work to one-loop order. In particular, the effective field theory analysis gives us confidence that our results will not change with the inclusion of two-loop and higher-order effects. Thus, we conclude that the right-handed neutrino mass scale plays no significant role in the determination of the Higgs spectrum in weak-scale supersymmetric models. where Q is the renormalization scale, ≡ 2 − 1 2 d and γ is Euler's constant. The O(m 2 D ) contribution to Σ hh (p 2 ) at leading order in the mass hierarchy [cf. Eq. (A.1)] is given by Next, we compute the O(m 2 D ) contributions [at leading order in the mass hierarchy, Eq. (A.1)] to the counterterm δ tan β in the various renormalization schemes. In the HM scheme, δ tan β is given by Eq. (2.26). Using the above expressions for the self-energy functions, along with Eq. (2.20) and the following tree-level relations (cf. Eq. (A.20) of Ref. [12]), Note that the O(m 2 D ) contributions to the counterterm δ tan β in the HM and DEC schemes are equivalent, in light of the absence of non-decoupling terms in Eq. (3.8). Indeed, the O(m 2 D ) contribution to δ tan β is independent of the tree-level Higgs mixing angle α. Although this result is obvious in the DEC scheme (which is defined via Higgs wave function counterterms that are evaluated at α = 0), the cancellation of the α-dependence in the O(m 2 D ) contribution to δ tan β HM [defined in Eq.(2.26)] is highly non-trivial.	2013-07-23T05:47:59.000Z	2013-04-22T00:00:00.000	{ "year": 2013, "sha1": "62ff659112abc760b0433c87dcd1544a5faa89cb", "oa_license": null, "oa_url": "http://arxiv.org/pdf/1304.6103", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "2079eeaa128d6c73c2bfe0e75027259f03ebfa00", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
238744198	pes2o/s2orc	v3-fos-license	Estimation and Inference of Extremal Quantile Treatment Effects for Heavy-Tailed Distributions Causal inference for extreme events has many potential applications in fields such as climate science, medicine and economics. We study the extremal quantile treatment effect of a binary treatment on a continuous, heavy-tailed outcome. Existing methods are limited to the case where the quantile of interest is within the range of the observations. For applications in risk assessment, however, the most relevant cases relate to extremal quantiles that go beyond the data range. We introduce an estimator of the extremal quantile treatment effect that relies on asymptotic tail approximation, and use a new causal Hill estimator for the extreme value indices of potential outcome distributions. We establish asymptotic normality of the estimators and propose a consistent variance estimator to achieve valid statistical inference. We illustrate the performance of our method in simulation studies, and apply it to a real data set to estimate the extremal quantile treatment effect of college education on wage. Introduction Quantifying causal effects of binary treatments on extreme events is an important problem in many fields of research. Examples include the effect of anthropogenic forcing on extreme precipitation, the effect of smoking on low birth weights, and the effect of education on high wages (e.g., Madakumbura et al., 2021;Dessì et al., 2018;Heckman et al., 2018). The quantile treatment effect (QTE) (e.g., Doksum, 1974;Lehmann and D'Abrera, 1975), which is based on the potential outcome framework, quantifies such causal effects. Formally, for a binary treatment D ∈ {0, 1} and an outcome Y ∈ R, let Y (0) and Y (1) denote the potential outcomes of Y under treatment D = 0 and D = 1 respectively. The fundamental problem of causal inference is that for each sample unit, only one of the outcomes can be observed, namely the one under the given treatment. The observed response is Y = Y (1)D + Y (0)(1 − D) (i.e., we make the stable unit treatment value assumption). Causal effects of D on Y can be defined in various ways according to different targets of interest. An example is the often used average treatment effect, which is defined as E[Y (1)] − E[Y (0)]. In this paper, the causal effect at extremely high/low quantiles is our main interest. Let the τ -QTE be defined as δ(τ ) := q 1 (τ ) − q 0 (τ ), where τ ∈ (0, 1) and q j (τ ) := inf{t ∈ R : F j (t) ≥ τ } denotes the τ -quantile of the potential outcome Y (j), and F j (t) denotes its distribution function. We will treat the case where τ is very close to 0 or 1. The QTE is generally not identifiable from observational data without making additional assumptions. A commonly made assumption, called the unconfoundedness assumption (e.g., Rosenbaum and Rubin, 1983), is that (Y (1), Y (0)) \|= D \| X for some set of observed covariates X ∈ R r . Under this assumption, the propensity score defined as Π(x) := P (D = 1 \| X = x) can be used to adjust for confounding in the binary treatment setting, and the QTE is identifiable. Along these lines, Firpo (2007) introduced an adjusted quantile estimator q j (τ ) (see (6)), defined as the minimizer of the inverse propensity score weighted empirical quantile loss, to estimate the τ -QTE in (1) for a fixed quantile level τ ∈ (0, 1). He also established the asymptotic normality of this estimator, allowing statistical inference. For extreme events, the interest lies in the τ -QTE for τ close to 0 or 1. Considering the lower tail, we allow the quantile τ = τ n to converge to zero as the sample size n tends to infinity. In particular, we distinguish between three cases based on different rates of τ n , and we call them: (a) intermediate: if τ n → 0 and nτ n → ∞; (b) moderately extreme: if τ n → 0 and nτ n → d > 0; and (c) extreme: if τ n → 0 and nτ n → 0. Here nτ n is the effective sample size or the expected number of observations below the τ n -quantile in a sample of size n. We note that the cases nτ n → d > 0 and nτ n → 0 are sometimes referred to as "extreme" and "very extreme" (Chernozhukov, 2005;Chernozhukov et al., 2016;Zhang, 2018). The asymptotic results of q j (τ ) with fixed τ in Firpo (2007) no longer hold in the framework of changing levels τ n , and Zhang (2018) first established the asymptotic theory for this estimator in this framework. Specifically, in the intermediate case, Zhang (2018) showed that the estimator is asymptotically normal and suggested a valid full-sample bootstrap confidence interval. For the moderately extreme case, he showed that the limiting distribution of the estimator is no longer Gaussian, and proposed a b out of n bootstrap for valid inference. For many applications, the most relevant quantiles are often those that go beyond the range of the data. For instance, in attribution studies, climate scientists investigate the causal effect of anthropogenic influences on climate extremes such as heavy precipitation. The quantiles of interest for such extreme events typically go far beyond the range of historical recordings and therefore extreme value extrapolation is required (e.g., Easterling et al., 2016;van Oldenborgh et al., 2017). Formally this corresponds to the case of extreme (rather than intermediate or moderately extreme) τ n -QTE where nτ n → 0. From now on, we use the notation p n to denote levels where p n → 0 and np n → d ≥ 0, which includes the extreme case, and we use τ n to denote only the intermediate levels where τ n → 0 and nτ n → ∞. To the best of our knowledge, there is no existing method for the estimation and inference of the p n -QTE where np n → 0. In particular, since the effective sample size below the p n -quantile tends to zero, the estimators q j (p n ) based on empirical quantile loss are no longer applicable. In this paper, we focus on heavy-tailed distributions, which have polynomially decaying tail probabilities and are thus heavier than Gaussian. Heavy tails are often encountered in risk analysis applications and many works therefore study this distribution class (e.g., Matthys et al., 2004;Wang et al., 2012;Athey et al., 2021;Xu et al., 2022). We propose a new quantile estimator Q j (p n ) for q j (p n ) based on parametric tail approximations from extreme value theory (de Haan and Ferreira, 2007), which enables us to extrapolate from intermediate to extreme quantiles. Indeed, based on the theory of regular variation, we have q j (p n ) ≈ q j (τ n ) (τ n /p n ) γ j , j = 0, 1, where γ j > 0 is the extreme value index of the potential outcome Y (j). Figure 1 in Section 2.1 illustrates the advantage of using extrapolation as in (2) for estimation of extreme quantiles compared to empirical estimates. Our proposed estimator Q j (p n ) (see (8) in Section 3) is obtained by plugging in the estimator q j (τ n ) (see (6)) from Firpo (2007) for intermediate quantiles and a newly proposed causal Hill estimator γ H j (see (7)) for the extreme value index based on inverse propensity score weighting. We use the Hill type instead of the Pickands type estimator for the extreme value index because the latter is known to suffer from high variance in the heavy-tailed case, but we would also like to note that the Hill type estimator is not invariant to location shift while the Pickands estimator is. The final p n -QTE estimator is δ(p n ) = Q 1 (p n ) − Q 0 (p n ), which we call the extremal QTE estimator. Beyond point estimation, we establish the asymptotic normality of this estimator. In particular, inspired by Zhang (2018) and Chernozhukov and Fernández-Val (2011), we propose a new normalizing factor for the extrapolation quantile estimators of two treatment groups, to deal with the problem that they may have different convergence rates. The asymptotic variance of the extremal QTE estimator is unknown, and we propose a technically tractable variance estimator. We prove that this estimator is consistent under an additional assumption (Assumption 6). Even when this assumption does not hold, this estimator is conservative, in the sense that it is still consistent to some quantity that is larger than the true variance. Thus it can be used to construct asymptotically honest confidence intervals for the extremal QTE. The extremal QTE estimator can be used for estimation and inference of moderately extreme and extreme quantiles for heavy-tailed distributions. It thus provides an alternative to Zhang (2018) for moderately extreme QTEs, and a first method for extreme QTEs. Our approach requires additional assumptions when compared to Zhang (2018), including most importantly the second-order regular variation, which is fairly standard in extreme value theory (e.g., de Haan and Ferreira, 2007). It is the price we need to pay in order to go to more extreme quantiles than Zhang (2018). The topic of causality for extreme events is receiving increasing interest. The line of work by ; ; Gnecco et al. (2021) and Mhalla et al. (2020) define structural causal models and investigate causal relationships in the setting where several variables are simultaneously extreme, and they focus on learning the unknown causal structure. In climate science, there is a large body of literature on attribution studies where the effect of climate change on weather extremes is analyzed (e.g., Hannart et al., 2016;Easterling et al., 2016;van Oldenborgh et al., 2017;Naveau et al., 2018Naveau et al., , 2020. These methods focus on model-based data where interventions on, say, carbon dioxide emissions, are possible, and no adjustment for confounding is required. Jana et al. (2021) propose a method to quantify the causal effect of London cycling superhighways on extreme traffic congestion, but no theoretical analysis of this method is done. Our method adds to this growing literature and provides a theoretically justified approach for estimation and inference of extremal QTEs in the presence of confounding variables. The paper is structured as follows. In Section 2, we review some key concepts from extreme value theory and the τ -QTE estimator. In Section 3, we propose the causal Hill extreme value index estimator and the extremal QTE estimator, and show their asymptotic normality. Furthermore, we propose a variance estimator and prove that it is consistent under an additional assumption, and that it is conservative otherwise. In Section 4, we present the finite sample behavior of our proposed extremal QTE estimator in different simulation settings, and compare it to existing methods. We apply our methodology to a real data set about college education and wage in Section 5. All proofs, more technical details and additional simulations can be found in the Supplementary Material. Any equations, theorems, etc., from the Supplementary Material are referred to starting with a letter A-F corresponding to the respective section in that document. To be in line with the literature on extreme value theory, we focus on extremal QTEs in the upper tail, that is, δ(1 − p n ). Results for the lower tail can be derived similarly. Extreme Value Theory We are interested in extreme quantiles with level p n where np n → d ≥ 0. Empirical estimates of extreme quantiles with d = 0 become highly biased and classical asymptotic theory no longer applies (see Figure 1). Extreme value theory studies methods for quantile extrapolation that result in more accurate estimates of extremal quantiles (e.g., de Haan and Ferreira, 2007). This theory relies on the following mild assumption on a distribution that guarantees that the tail can be well approximated in a parametric way. Let the random variable Y have distribution F and quantile function q(·) = F ← (·), where f ← (x) := inf{y ∈ R : f (y) ≥ x} is the left-continuous inverse of a non-decreasing function f . Definition 1. (cf. de Haan and Ferreira (2007)) The distribution F is in the max-domain of attraction of a generalized extreme value distribution if there exist γ ∈ R and sequences of constants a n > 0 and b n ∈ R, n = 1, 2, . . ., such that for all x such that 1 + γx > 0. For γ = 0, the right hand side is interpreted as exp(−e −x ). The parameter γ is called the extreme value index (EVI). This condition is mild as it is satisfied for most standard distributions, for example, the normal, Student-t and beta distributions. For a complete characterization of the max-domain of attraction of the three regimes (γ < 0, γ = 0, γ > 0), we refer to Resnick (2008) and Embrechts et al. (1997). We focus on the heavy-tailed case where γ > 0, that is, distributions F with regularly varying tails 1 − F (x) = L(x)x −1/γ , where L is a slowly varying function at ∞. The regular variation of the tail of F can be equivalently expressed in terms of the tail quantile function U := (1/(1 − F )) ← . The max-domain of attraction condition (3) for γ > 0 is equivalent to Relation (4) and the fact that the quantile function q = F ← can be expressed as q(τ ) = U (1/(1 − τ )) for τ ∈ (0, 1), imply that for large enough n, where τ n is a sequence such that τ n → 0 and τ n > p n . By using an intermediate sequence τ n , relation (5) allows us to extrapolate from intermediate to extreme quantiles. We illustrate the benefit of using the extreme value extrapolation (5) in Figure 1. It can be seen that the empirical estimates are strongly biased for extreme quantiles when the effective sample size np n < 1, while the extrapolation allows for approximately unbiased estimates. Figure 1: Averaged quantile estimates (over 1000 repetitions) at different probability levels 1−p n fitted on n = 1000 i.i.d. samples from a Fréchet distribution with shape 3 (i.e., γ = 1/3), location 0 and scale 1. The corresponding effective sample size np n is shown with parentheses. The solid black line denotes the true quantiles, the dotted red line denotes the empirical estimator, and the dotted green line denotes the extrapolation method (5), where the intermediate quantile level is 1 − τ n = 90% and the EVI γ is estimated by the Hill estimator (Hill, 1975). To analyze the asymptotic distribution of the extrapolated quantiles, usually a second-order condition is assumed. Definition 2. (cf. de Haan and Ferreira (2007)) The function U is of second-order regular variation with first-order parameter γ > 0 and second-order parameter ρ ≤ 0 if there exists a positive or negative function A with lim t→∞ A(t) = 0 such that for all x > 0 The function A is called the second-order auxiliary function. For ρ = 0, x ρ −1 ρ is defined as lim ρ→0 We note that Definition 2 is stronger than Definition 1: if U is of second-order regular variation with first-order parameter γ > 0, then F satisfies the max-domain of attraction condition with extreme value index γ. Second-order regular variation is satisfied by many popular distributions, and it is often possible to derive the function A and the value of ρ explicitly; see Alves et al. (2007) for details and examples. Estimation of the τ -QTE The estimation of τ -QTE from Firpo (2007) relies on the propensity score Π(x), which is used to adjust for confounding in the binary treatment setting. In particular, Firpo (2007, Corollary 1) shows that the QTE is identifiable under the following assumptions. ii) X has compact support Supp(X) and there exists c > 0 such that c < Π(x) < 1 − c for all x ∈ Supp(X). Assumption 1 i) is the unconfoundedness condition and Assumption 1 ii) is called the common support assumption. Both assumptions are fairly standard in causal inference literature, and we impose them throughout this paper. The propensity score Π(x) is generally unknown in practice and needs to be estimated. For n independent copies (Y i , D i , X i ) n i=1 of (Y, D, X), we follow Hirano et al. (2003), Firpo (2007) andZhang (2018) and use the nonparametric sieve method to obtain an estimate Π(x) (see Section A in the Supplementary Material for more details). Based on inverse propensity score weighting, Firpo (2007) proposed the following estimators for the quantiles of the potential outcomes Y (1) and Y (0): The τ -QTE is then estimated by q 1 (τ ) − q 0 (τ ). Under Assumption 2 below and the two regularity Assumptions 7 and 8 in Section B of the Supplementary Material, Zhang (Theorem 3.1 2018) showed that for the intermediate quantile index τ n (i.e., τ n → 0, nτ n → ∞), the τ n -QTE estimator q 1 (1 − τ n ) − q 0 (1 − τ n ) is asymptotic normal. For the moderately extreme case (i.e., τ n → 0 and nτ n → d > 0), however, Zhang (2018) showed that the limiting distribution is no longer Gaussian. Assumption 2. (Regularity conditions on the potential outcome distributions) For j = 0, 1: i) Y (j) and Y (j) \| X are continuously distributed with densities f j and f j\|X , respectively. ii) The density f j of Y (j) is monotone in its upper tail. iii) The distribution function F j of Y (j) belongs to the max-domain of attraction of a generalized extreme value distribution with extreme value index γ j (see Definition 1). Extremal Quantile Treatment Effect Estimation for Heavy-Tailed Models We now go beyond the intermediate and moderately extreme cases and propose an extremal QTE estimator based on quantile extrapolation, that can be used in moderately extreme and extreme cases (i.e., p n → 0, np n → d ≥ 0). We also derive its asymptotic normality and propose an asymptotic variance estimator which enables us to construct a confidence interval for the extremal QTE. Extremal QTE Estimator Our extremal QTE estimator is based on the quantile extrapolation approach (see approximation (5)), so estimators for the EVIs of the potential outcome distributions are required. In the classical setting, the Hill estimator introduced by Hill (1975) is a common choice for heavy-tailed cases. In the potential outcome setting, however, the classical Hill estimator is not appropriate due to confounding. Therefore, we propose the following causal Hill estimators that adjust for confounding via the estimated propensity score Π (see (15)): Zhang (2018) also proposed two EVI estimators for short-tailed distributions in his supplementary material, one is the Pickands type estimator and one is the Hill type estimator. Rather than extrapolation, his goal of using the EVI estimator is to estimate the 0-th QTE, that is, the lower endpoint of the distribution. His Hill type estimator is moment-based, which also uses inverse propensity score weighting, but with the true propensity score. In the simulations in Section 4, we implement the quantile extrapolation approach with his Pickands type estimator and compare it to our proposed estimator (7). Building on the causal Hill estimators γ H j in (7) and the intermediate quantile estimator (6), we propose the following quantile extrapolation estimator for j = 0, 1, and the final extremal QTE estimator is defined as the difference of the extrapolated quantiles: For simplicity, we use the same intermediate level τ n for both extrapolation estimators Q 1 (1−p n ) and Q 0 (1 − p n ) in this paper, but in principle, one can use different intermediate levels for each potential outcome. The latter might be advantageous if the potential outcome distributions have very different tail behaviors, or if there is a severe imbalance between treated and non-treated samples. To obtain the extremal QTE estimator (9), the intermediate level τ n (or k if we consider τ n = k/n) needs to be chosen. The optimal choice of the τ n depends on the underlying data distribution and is difficult in practice, and there is a bias-variance trade-off. Specifically, if τ n is too small, then the effective sample size nτ n is small and this will lead to high variance. On the other hand, if τ n is too large, then the assumptions of extreme value theory may not hold because we are no longer in the tail of the distribution, and this will lead to high bias. In practice, there are some commonly used approaches for choosing τ n . The simplest one is to set τ n to some reasonable fixed value based on the background knowledge of the concrete problem. One can also plot the estimates depending on different τ n and then select τ n in the first stable region of the plot (e.g., Resnick, 2007). There are also adaptive methods to approximate the optimal τ n (e.g., Drees and Kaufmann, 1998;Boucheron and Thomas, 2015). Asymptotic Properties The main theoretical result in this subsection is Theorem 3, which shows the asymptotic normality of the extremal QTE estimator δ(1 − p n ). The major steps to prove this theorem are the following: (1) showing that the asymptotic behavior of the quantile extrapolation estimator Q j (1 − p n ) depends only on the asymptotic distribution of the causal Hill estimator γ H j (see Lemma 2); (2) deriving the asymptotic distribution of γ H j (see Theorem 2), which builds on an asymptotic linearity result (see Theorem 1); (3) introducing a new normalizing factor β n (see formula (10)) when deriving the asymptotic distribution of δ(1 − p n ) to account for the issue that Q 1 (1 − p n ) and Q 0 (1 − p n ) may have different convergence rates. Asymptotic Properties of the Causal Hill Estimator We first present a result showing that under the same conditions as the Theorem 3.1 of Zhang (2018), our proposed causal Hill estimator is consistent. Lemma 1. Suppose that Assumptions 1, 2, 7 and 8 hold, and nτ n → ∞ and τ n → 0. If for j = 0, 1, the extreme value index γ j > 0, then To obtain asymptotic normality of the causal Hill estimator, we require Assumption 3 below, which is a second-order regular variation assumption. This assumption is standard to obtain asymptotic normality results for heavy-tailed distributions, and it is satisfied by most heavy-tailed distributions such as the Student-t and the Fréchet distribution (e.g., de Haan and Ferreira, 2007). Assumption 3. For j = 0, 1, the tail function U j = (1/(1 − F j )) ← of Y (j) is of second-order regular variation (see Definition 2) with extreme value index γ j > 0, second-order auxiliary function A j and second-order parameter ρ j ≤ 0. To guarantee that the estimated propensity score is compatible with the causal Hill estimator, we also require Assumption 4 below, which is a regularity assumption on the sieve estimator. This assumption is of similar type as Assumption 7 which is used in the Theorem 3.1 of Zhang (2018). Assumption 4. x with all derivatives bounded by some N n uniformly over Supp(X). Here n τn N n h −t/2r n → 0 as n → ∞, where h n is the number of sieve bases and r is the dimension of X. Before showing asymptotic normality of the causal Hill estimator, we show an important asymptotic linearity result under the above two extra assumptions. Theorem 1. Suppose that Assumptions 1 -4, 7 and 8 hold, and k = nτ n → ∞ and τ n = k/n → 0. If for j = 0, 1, √ kA j (n/k) → λ j ∈ R and the extreme value index γ j > 0, then Remark 1. φ i,j,n is the influence function for the intermediate quantile estimator q j (1 − τ n ) which also appears in Zhang (2018) (see also Theorem 5), and ψ i,j,n is the new influence function appearing in our result, which has a similar form as φ i,j,n . The factors E [S i,j,n \| X i ] and E [T i,j,n \| X i ] in the influence functions correspond to the information gain from the nonparametric estimation of the propensity score. Zhang (2018) made the observation that since negligible under suitable integrability conditions. The same holds for the information gain term in ψ i,j,n because E [S i,j,n \| X i ] is also of order O p (τ n ) (see Lemma 9). We discuss this in more detail when considering variance estimation in Section 3.3. Given the asymptotic linearity result in Theorem 1 and the fact that the influence functions depend on the sample size n, we will use a triangular array central limit theorem to obtain asymptotic normality. This requires that the covariance matrix Σ n of the random vectors (ψ i,1,n , ψ i,0,n , φ i,1,n , φ i,0,n ) converge (see Assumptions 8 and 9). The extra Assumption 9 is of similar flavor as Assumption 8 used in Theorem 3.1 of Zhang (2018). In fact, we can show that the sequence Σ n is bounded by using similar arguments as in the proof of Theorem 2. Its convergence is therefore mostly a technical condition. We now give the asymptotic normality result of the causal Hill estimator. Asymptotic Properties of the Extremal QTE Estimator We now study the asymptotic properties of the quantile extrapolation estimator Q j (1−p n ) in (8). We first give the following lemma which shows that the asymptotic behavior of Q j (1 − p n ) only depends on the asymptotic distribution of the EVI estimator. Lemma 2. Suppose that Assumptions 1 -4 and 7 -9 hold, and k = nτ n → ∞, k/n → 0, np n = o(k), and log(np n ) = o( √ k). If for j = 0, 1, √ kA j (n/k) → λ j ∈ R and the extreme value index γ j > 0, then In particular, the above implies that Lemma 2 (and the main result Theorem 3) allows that np n → 0, but it cannot converge to zero arbitrarily fast as log(np n ) = o( √ k). This is reasonable since it means that there are limitations on how far the extrapolation can be pushed. The other rate condition np n = o(k) is also natural since we are interested in the case where p n converges to 0 much faster than τ n . These rate conditions are standard (see, e.g., de Haan and Ferreira, 2007). We already know from Theorem 1 that the causal Hill estimator is asymptotically normal. Thus, to show asymptotic normality of the extremal QTE estimator δ(1 − p n ) in (9), the only remaining difficulty is that Q 1 (1 − p n ) and Q 0 (1 − p n ) can have different normalizing factors and convergence rates. This is problematic if the ratio of the normalizing factors oscillates. Zhang (2018) encountered the same issue, and he followed the idea from Chernozhukov and Fernández-Val (2011) to construct a feasible normalizing factor under the assumption that the ratio of normalizing factors converges. We proceed similarly. Specifically, based on Lemma 2, we introduce the following normalizing factor and make the following assumption. Assumption 5 states that the tails of the potential outcome distributions are either comparable or that one of them is heavier than the other. This is a fairly standard assumption satisfied by many models. Using the normalizing factor (10), we can show asymptotic normality of the extremal QTE estimator. Theorem 3. Suppose that Assumptions 1 -5 and 7 -9 hold, and k = nτ n → ∞, k/n → 0, np n = o(k) and log(np n ) = o( √ k). If for j = 0, 1, √ kA j (n/k) → λ j ∈ R and the extreme value index γ j > 0, then where µ = v T κ w λ,ρ and σ 2 = v T κ BΣB T v κ with B and Σ defined as in Theorem 2, and Due to the asymptotic bias of γ H j (see Theorem 2), there is also an asymptotic bias of the extremal QTE estimator δ(1 − p n ), which affects the validity of our later proposed confidence interval (see (12)). This asymptotic bias equals 0 if √ kA j (n/k) → 0. Recall that lim t→∞ A j (t) = 0 by the second-order regular variation assumption (see Assumption 3), so √ kA j (n/k) → 0 holds if k grows not too fast. The rate at which A j (n/k) tends to zero is unknown, and we therefore advise being conservative in the sense that one should choose k (or equivalently, τ n ) rather small in practice to ensure that √ kA j (n/k) → 0, and hence the asymptotic bias is negligible. Variance Estimation and Confidence Intervals In order to conduct statistical inference based on Theorem 3, a consistent estimator of the asymptotic variance σ 2 is needed. The main difficulty in estimating σ 2 lies in estimating the corresponding matrix Σ, which is the limit of the covariance matrices of the influence functions. Recall from Remark 1 that these influence functions contain terms describing the information gain from nonparametric estimation of the propensity score. Firpo (2007) encountered a similar issue and he proposed a nonparametric regression approach to estimate the contribution of the information gain to the variance. In this paper, however, we will not go in this direction. Instead, we show that under suitable assumptions, the information gain for the proposed extremal QTE estimator is actually negligible, which can simplify the covariance matrix needed to be estimated, and thus a simpler and computational cheaper method can be proposed. Specifically, we require the following assumption: Lemma 9 in the Supplementary Material shows that both P (Y (j) > q j (1 − τ n ) \| X) and E [S j,n \| X] are of order O p (τ n ). Hence Assumption 6 holds under suitable integrability conditions, and Section C in the Supplementary Material presents a concrete example where it is satisfied. We propose the following variance estimator where and Σ is defined in (24) and its entries are estimated using inverse propensity score weighting; see Section D of the Supplementary Material for the for details. The following result shows that this estimator is consistent. Based on Theorem 3 and the variance estimator (11), we propose the following approximate (1 − α)−confidence interval of the extremal QTE: where β n is the normalization constant in (10) and Since Assumption 6 only affects the information gain terms in the covariance matrix Σ and these terms reduce the variance, even Assumption 6 does not hold, the variance estimator is conservative, in the sense that it is still consistent to some quantity σ 2 (see (22) and the proof of Theorem 4) that is larger than the true variance σ 2 ; see Lemma 3 below. Therefore, even if Assumption 6 does not hold, it is safe to use our estimator σ 2 . Simulations We conducted simulations to examine the finite sample behavior of our proposed extremal QTE estimator (9) and the related confidence interval (12), and to compare them to other methods. All simulations were carried out in R and the code is available at Github. Simulation Set-up Throughout the simulation study, we consider univariate covariate X as Zhang (2018). Specifically, let X and U be uniformly distributed random variables on [0, 1] and assign the treatment by D = 1 U ≤Π(X) with propensity score Π(x) = 0.5x 2 + 0.25. We generate the outcomes from the following three models: where S follows a Student-t distribution with 3 degrees of freedom, C s is Fréchet distributed with shape parameter s, location 0 and scale 1, and P a,b is Pareto distributed with shape parameter a and scale b. The EVIs of the potential outcome distributions are γ 1 = γ 0 = 1/3 for model H 1 and γ 1 = 1/2 and γ 0 = 1/3 for model H 2 . For model H 3 , a small calculation yields γ 1 = γ 0 = 4/7. Models H 2 and H 3 are more heavy-tailed than H 1 . We consider data sets with sample size n ∈ {1000, 2000, 5000} and aim to estimate the (1 − p n )-QTE with p n ∈ {5/n, 1/n, 5/(n log n)}. Throughout, the target coverage for the confidence intervals is 90% (i.e., α = 0.1). For all bootstrap based methods, we use 1000 bootstrapped data sets. The empirical squared error and coverage are calculated based on 1000 sampled data sets. Implemented Methods For point estimation of the extremal QTE, we compare the squared errors of three methods: where the quantile estimators are defined by (6). It was proposed by Firpo (2007) and further studied by Zhang (2018). • Extrapolation with a causal Pickands estimator: where This estimator is based on quantile extrapolation with the causal Pickands EVI estimator γ P j proposed in the supplementary materials of Zhang (2018). • Extremal QTE estimator (see (9)): our proposed estimator based on quantile extrapolation with the causal Hill EVI estimator. For the confidence interval of the extremal QTE, we compare the empirical coverages of the following four methods: • Zhang: the b out of n bootstrap confidence interval proposed by Zhang (2018) that builds on the Firpo-Zhang estimator. We use the "with replacement" version as Zhang (2018) suggested in his paper. Its tuning parameters are described in Section E.1 of the Supplementary Material. • BS Pickands: a bootstrap based method with the bootstrap confidence interval where δ (1−p n ) is the point estimate (13) of the extremal QTE based on the full sample and σ * is the estimated standard deviation of this estimate via the non-parametric bootstrap. • BS Hill: a non-parametric bootstrap based method as (14), but using (9) to obtain the point estimate. • Extremal QTE CI: our proposed confidence interval (12). The whiskers of the box plots correspond to the 0.1 and 0.9 quantiles, the black horizontal line in the box corresponds to the median, and the square indicates the mean. Please note that the log-scale is used for the y-axis. For the intermediate quantile level τ n = k/n of the extrapolation based methods, we use k = n 0.65 . This value guarantees that all rate assumptions about k, n and p n in Theorems 3 and 4 are satisfied. An additional sensitivity analysis can be found in Section E.3 of the Supplementary Material. We note that k = n 0.65 may not be optimal in all settings, and we use it in our simulations mostly for convenience. Choosing the optimal data-dependent τ n is a difficult problem in extreme value theory. In practice, we recommend choosing it by plotting the estimates using different τ n and selecting the τ n in the first stable region of the plot (e.g., Resnick 2007). Please also see the real data application in Section 5 for an illustration of this approach. For the size of sieve basis functions h n in the propensity score estimation, we use h n = 2n 1/11 . Note that this choice is only for the case of univariate X, and please see Section B of the Supplementary Material for some justifications about this choice. Specifically, we have h 1000 = h 2000 = 3 and h 5000 = 4. In practice, people may choose the sieve basis functions according to their specific problem and use model selection methods such as cross-validation. Please see the real data application in Section 5 for an illustration. Simulation Results The squared errors of the point estimates are shown in Figure 2. We see that our proposed extremal QTE estimator generally performs better than the other two methods. In particular, it exhibits the lowest mean squared error (MSE) over almost all settings. This is especially true for the more heavy-tailed models H 2 and H 3 , in which our method greatly outperforms the others. The extrapolation based method using the Pickands EVI estimator has the worst performance, which is not surprising as the Pickands estimator is known to suffer from high variance in heavy-tailed settings. This also indicates that choosing a suitable EVI estimator is crucial for the extrapolation based method. Figure 3: Coverage of different methods to construct confidence intervals. The target coverage is 90% and is indicated by the solid horizontal black lines. The dots indicate the empirical coverage over 1000 simulations, and the error bars indicate an approximate normal based 95%-confidence interval for the true coverage over 1000 simulations. Figure 3 compares the empirical coverage of the different confidence intervals. We see that "Zhang" performs quite well for the not so extreme quantile level p n = 5/n, but that it can undercover largely when the quantile index becomes more extreme. Such results were expected as this method is designed for the moderately extreme case where np n → d > 0, and not for the extreme case where np n → 0. The bias of the Firpo-Zhang estimator in the extreme case could be a reason for this undercoverage. We see that "BS Pickands" may suffer from both undercoverage (e.g., setting p n = 5/n of H 1 ) and overcoverage (e.g., setting p n = 5/n of H 2 ). In comparison, "BS Hill" and our proposed confidence interval "Extremal QTE CI" perform better, with empirical coverage close to the nominal level. The performance of "BS Hill" shows that the bootstrap based method may be valid, and it would be interesting to formalize this in future research. Compared to "BS Hill", "Extremal QTE CI" has computational advantage, as it does not require bootstrapping the data. The asymptotic normality result in Theorem 3 is also confirmed by the normal Q-Q plot in Section E.2 of the Supplementary Material. Extremal Quantile Treatment Effect of College Education on Wages The causal effect of education on wage has been studied extensively in the literature (e.g., Card, 1995;Heckman and Vytlacil, 1998;Card, 1999;Messinis, 2013;Heckman et al., 2018). It is wellknown that wage exhibits heavy-tailed behavior, and there can be considerable confounding between education and wage (e.g., Griliches, 1977;Heckman et al., 2006). In this section, we apply our method to obtain point estimates and confidence intervals of the extremal QTE of college education on wage in the upper tail of the distribution, where we focus on the 0.99, 0.995, 0.997 and 0.999-QTEs. As comparison, we also implement the Firpo-Zhang estimator and the b out of n bootstrap confidence interval of Zhang (2018). We use data from the National Longitudinal Survey of Youth (NLSY79). It consists of a representative sample of young Americans who were between 14 and 21 years old at the time of the first interview in 1979, and contains a wide range of information about education, adult income, parental background, test scores and behavioral measures of the study participants. In particular, we use the NLSY79 data analyzed by Heckman et al. (2006Heckman et al. ( , 2018, which is available from https://www.journals.uchicago.edu/doi/suppl/10.1086/698760. This data set consists of male participants who finished their education before the age of 30 and were not in the military. We only consider participants that graduated from high school. The same (or similar) data set was also used in other literature (e.g., Brand and Xie, 2010;Cheng et al., 2021;Zhou, 2022). The outcome Y is the hourly wage (in US dollar) at age 30, and the treatment D equals 0 if the person did not receive any college education, and 1 otherwise. For the covariates X used for propensity score estimation, we follow Heckman et al. (2018) and consider race, region of residence in 1979, urban status in 1979, broken home statue, age in 1979, number of siblings, family income in 1979, education (highest grade completed) of father and mother, scores from the Armed Services Vocational Aptitude Battery (ASVAB) test, and GPAs from 9th grade core subjects (language, math, science and social science). This leads to 19 covariates in total as some of the variables are categorical. We omit samples with missing values, leading to a data set with n = 805 samples, among which 432 (53.7%) went to college. In this data set, the 0.99, 0.995, 0.997 and 0.999 quantiles of hourly wage are 50.47, 60.72, 85.50 and 154.73 US dollar, respectively. For the propensity score estimation, considering that there are 19 covariates and the sample size is 805, we refrain from using too many high-order terms in the sieve method to avoid overfitting. In particular, we consider two approaches. The first approach, which we refer to as PROP1, uses only linear terms, leading to sieve basis functions H hn (x) = (H hn,j (x)) j=1,...,hn = (1, x 1 , . . . , x 19 ) with h n = 20. This approach is equivalent to logistic regression, which is widely used in practice for the propensity score estimation, and is the default option in many packages (Olmos and Govindasamy, 2015). The second approach, which we refer to as PROP2, allows second-order terms and uses model selection to avoid overfitting. Specifically, we first apply a model selection procedure on the 19 covariates. Then, we do another round of model selection, allowing only first-and second-order terms of all covariates that were selected in the first step. Both model selection steps are implemented using the R package glmulti (Calcagno and de Mazancourt, 2010) with Akaike's information criteria and a genetic search algorithm. The resulting model of PROP2 can be found in Section F.1 of the Supplementary Material. We use the same estimated propensity scores for all methods, and we mention that with a larger sample size, one may consider to use more higher-order terms in the sieve method for the propensity score estimation. To select the tuning parameter k for our method, here we use the approach of plotting the estimated EVI and QTE versus k and then choose k from the first stable region (e.g., Resnick, 2007). The corresponding plots can be found in Section F.2 of the Supplementary Material. Based on these plots, we choose k = 85. Note that the choice used in Section 4 leads to k = 805 0.65 ≈ 77, resulting in similar confidence intervals as using k = 85. Figure 4 presents the results of the different methods. Considering the point estimate, we see that both Firpo-Zhang and our method give positive QTE estimates, but the corresponding values are quite different in some cases. In particular, the estimated values of Firpo-Zhang are not monotonically increasing when the quantile levels become more extreme, whereas ours are monotonic for these data. Such monotonicity implies that college education would have a stronger effect on wages for higher quantiles, which seems possible. The confidence intervals of both methods mostly lie on the positive part, showing strong evidence that the QTEs are positive. The intervals of our method are considerably narrower than Zhang's b out of n bootstrap intervals for the 0.997-and 0.999-QTEs, a clear advantage of our methodology. We also observe that the propensity scores estimated by the second approach lead to wider confidence intervals in all cases. At last, we would like to note that the unconfoundedness assumption can not be verified in practice. For example, one may suspect that cognitive ability is not explicitly controlled for in this data example, so the unconfoundedness assumption may not hold. But since we have controlled for many important related covariates, we think that the unconfoundedness assumption is still a suitable approximation. For more discussion we refer to Brand and Xie (2010). Discussion We propose a method to estimate the extremal QTE of a binary treatment on a continuous outcome for heavy-tailed distributions under the unconfoundedness assumption. Our method, which we call the extremal QTE estimator, builds on the quantile extrapolation approach from extreme value theory. We use the inverse propensity score weighted intermediate quantile estimates of Firpo (2007) and our newly proposed causal Hill estimator to extrapolate to extreme quantiles. We show the asymptotic normality of the causal Hill estimator and the extremal QTE estimator. In particular, asymptotic normality of the extremal QTE estimator holds for extremal (1 − p n )-QTEs, where np n may converge to 0. This is particularly important since it represents a common setting in risk assessment where the quantities of interest are beyond the range of the data. To the best of our knowledge, our approach is the first that achieves this. We also develop an estimator for the asymptotic variance which is consistent under suitable assumptions. This enables us to construct confidence intervals for the extremal QTEs. Simulations show that our method generally performs well. As mentioned before, there is an asymptotic bias term of our proposed extremal QTE estimator δ(1 − p n ) (see Theorem 3), which is due to the asymptotic bias of the causal Hill estimator γ H j . In this paper, we suggest choosing a sufficiently small k so that the asymptotic bias is negligible. It would be interesting and desired to formally propose bias-corrected versions of γ H j and δ(1 − p n ) in future research. One potential issue of introducing inverse propensity score weighting to EVI estimators is that it complicates the estimation of the asymptotic variance, making statistical inference difficult. Bootstrap methods can be useful in practice for constructing confidence intervals for QTEs, as our simulation results suggest. It is important to study the theoretical validity of such bootstrap based methods in future research. In particular, it would be interesting to investigate whether the bootstrap based methods are valid even without Assumption 6. The proposed method is just the first step towards the goal of doing causal inference for extremes. The considered causal inference setting is the most common and simplest one: binary treatment with assumed unconfoundedness. We believe that it is possible to extend it in many ways to fit a range of applications. For example, it would be interesting to generalize our method to categorical and continuous treatment settings by using the generalized propensity score (Imbens, 2000). One may also consider extending our method to other causal inference settings, such as instrumental variable settings that allow for some types of confounding. At last, quantile extrapolation is not limited to heavy-tailed distributions, and it would be interesting to extend our proposed extremal QTE estimator to other settings where the potential outcome distributions may have lighter tails. A Details of the estimated propensity score using sieve method Suppose that we observe n independent copies (Y i , D i , X i ) n i=1 of (Y, D, X). The main idea of the sieve method is to approximate the logit of the propensity score by a linear combination of sieve basis functions, and then estimate the propensity score by where H hn = (H hn,j ) j=1,...,hn : R r → R hn is a vector consisting of sieve basis functions, and π n := arg max and L(a) := 1/(1 + e −a ) is the sigmoid function. Let H hn = (H hn,j ) j=1,...,hn : R r → R hn be a vector consisting of h n sieve basis functions. Following Hirano et al. (2003) and Firpo (2007), we use polynomials as the sieve basis functions in this paper. In particular, we require that H hn,1 = 1 and for all m such that h n > (m+1) r , the span of H hn contains all polynomials up to order m. For an illustration purpose, some possible examples for H hn are H hn (x) = (1, x 1 , x 2 , . . . , x r ) or H hn (x) = (1, x 1 , x 2 , . . . , x r , x 2 1 , x 2 2 , . . . , x 2 r ). The crucial point in sieve estimation is that the dimension of the sieve space h n grows to infinity at an appropriate speed with the sample size n. In other words, with larger sample size, one may consider a more complex model for the estimation. B Regularity assumptions for sieve estimation and the central limit theorem For sieve estimation, we require certain regularity assumptions: Assumption 7. i) X is continuous and has density f X such that ∃c > 0 : c < f X (x) < 1 c , ∀x ∈ Supp(X). ii) Π(x) is s-times continuously differentiable with all the derivatives bounded, where s ≥ 4r and r denotes the dimension of X. iii) E[τ n − 1 Y (j)>q j (1−τn) \|x] is u-times continuously differentiable in x with all derivatives bounded by some M n uniformly over Supp(X), where u ∈ N. In Assumption 7, i) and ii) are standard assumptions for sieve estimation. iii) and iv) were introduced by Zhang (2018) for the intermediate quantile estimation, and we refer to Zhang (2018) for more discussions about these two assumptions. Newey (1994) showed that if H hn consists of orthonormal polynomials, then ζ(h n ) = O(h n ). In this case, if h n = c 2 n c 1 for some positive constants c 1 , c 2 , Assumption 7 iv) is equivalent to c 1 < 1 6 , τ n n 11c 1 −1 → 0, τ n n c 1 (6−s/r)+1 → 0 and M n n 1−c 1 u/r /τ n → 0. In particular, since τ n → 0 and nτ n → ∞, this assumption holds if we have c 1 ≤ 1 11 and sufficient smoothness. Below we present the regularity assumptions for the central limit theorem. 1 For any vector or matrix A, the norm A = tr(A T A). C Examples satisfying Assumption 6 Assumption 6 is sarisfied for the following random scale model in Example 1. Example 1. Let X ∈ R r be a random vector and let h j : R r → R + , j = 0, 1, be functions satisfying C 1 ≤ h j (X) ≤ C 2 almost surely for some real numbers C 1 , C 2 such that 0 < C 1 ≤ C 2 < ∞. Let ε 0 , ε 1 be random variables independent of X and let potential outcome Let F ε j be the distribution function of ε j . Suppose that U ε j := (1/(1−F ε j )) ← satisfies the secondorder regular variation condition with extreme value index γ j > 0 and second-order parameter ρ j ≤ 0. Then U j = (1/(1 − F j )) ← satisfies the second-order regular variation condition with the same parameters γ j , ρ j , and Assumption 6 is met. We give two concrete examples for the distribution of j in Example 1. Example 2. The first example is the Student t-distribution. Let F ν be the CDF of the Student t-distribution with ν degrees of freedom. It is known that F ν satisfies the second-order regularvariation condition with first-order parameter −ν and second-order parameter ρ = −2 (see e.g. Example 3 in Hua and Joe (2011)). Therefore, by Theorem 2.3.9 in de Haan and Ferreira (2007), U ν := (1/(1 − F ν )) ← is of second-order regular variation with extreme value index γ = 1/ν and second-order parameter ρ = −2. One special case is the Cauchy distribution for ν = 1. We prove the following Lemma 4. The claim of Example 1 then follows from setting t = q j (1 − τ n ). Lemma 4. Let X ∈ R r be a random vector and let h : R r → R + be a function satisfying C 1 ≤ h(X) ≤ C 2 almost surely for some real numbers C 1 , C 2 such that 0 < C 1 ≤ C 2 < ∞. Let ε be a random variable such that ε \|= X and let Z = h(X) · ε. Denote the CDFs of Z and ε by F and F , respectively. Suppose that U = (1/(1 − F )) ← is of second-order regular variation with extreme value index γ > 0 and second-order parameter ρ ≤ 0. Then, U = (1/(1 − F )) ← is of second-order regular variation with extreme value index γ and second-order parameter ρ. In addition, for t → ∞, we have and Proof of Lemma 4. First, since replacing ε by ε1 ε≥0 does not change the CDF of ε or Z on (0, +∞), we can assume without loss of generality that ε ≥ 0. By Theorem 2.3.9 in de Haan and Ferreira (2007), U being second-order regular variation with extreme value index γ > 0 and second-order parameter ρ ≤ 0 is equivalent to 1 − F being second-order regular variation with first-order parameter −1/γ and second-order parameter ρ. Therefore, equation (25) in Hua and Joe (2011) implies that 1 − F is of second-order regular variation with parameters −1/γ and ρ, given the condition that there exists δ > 0 such that E h(X) 1/γ−ρ+δ < ∞, The required condition holds in our case because E h(X) 1/γ−ρ+δ ≤ C 1/γ−ρ+δ 2 < ∞ for all δ > 0. Theorem 2.3.9 in de Haan and Ferreira (2007) then implies that U is second-order regularly varying with parameters γ and ρ. Now we prove claim (17). First, since ε \|= X, we have F (t) = P (ε ≤ t/h(X)) = E F (t/h(X)) , and thus for all t > 0, Because 1 − F is of second-order regular variation with first-order parameter −1/γ, we have Since Now we prove claim (18). We have that almost surely, For the first term, since F (ε) is uniformly distributed, we have that log follows a standard exponential distribution. Thus Based on (19), we have that log((1 − F (t))/(1 − F (t/C 2 ))) is of order O(1), and thus E log For the second term, we have Since 1 − F is of second-order regular variation, we have that 1 ≤ 1− F (q) 1− F (q·C 2 /C 1 ) = O(1), and it follows that Combining this with (19) yields Therefore, by taking squares and expectations on both sides of (20), and using the above two results, we have which proves claim (18). D Details of Variance Estimation The true variance in Theorem 3 is σ 2 = v T κ BΣB T v κ , where we denote the true covariance matrix where H 1 , H 0 , H 10 , H 10 are defined according to Assumption 8. and G 1 , G 0 , G 10 , J 1 , J 0 , J 10 , J 01 are defined according to Assumption 9. Let with the simplified covariance matrix where the entries are defined as the following limits: Under Assumption 6, the true covariance matrix Σ is simplified to Σ (see Lemma 10), which leads to the estimator with entries where q j (1 − τ n ) is the estimator defined by (6), and Π is the estimated propensity score in (15). Finally, the estimator of the variance is given by Figure 5: Normal Q-Q plots of QTE estimates from different methods for model H 2 with n = 5000, p n = 5/(n log(n)). The x-axis corresponds to the theoretical quantiles of the standard normal distribution, and the y-axis corresponds to the sample quantiles of the QTE estimators. For the spacing parameter m and τ n,0 in the feasible normalizing factor, we use the formulas described in Section 5.5 of Zhang (2018): τ n,0 = min 10 n , 0.1b n and m = 1 + 10 nτ n,0 . E.2 Q-Q plots To empirically verify the asymptotic normality result of our extremal QTE estimator in Theorem 3, we show in Figure 5 its related normal Q-Q plot. As comparison, we also present the normal Q-Q plots for the extrapolation estimator with the causal Pickands estimator and the Firpo-Zhang estimator. The Q-Q plots of all settings are similar, thus we only present those of model H 2 with n = 5000 and p n = 5/(n log(n)) as an example. From the plot, our proposed extremal QTE estimator is approximately normal, which empirically verifies Theorem 3. The other two estimators, however, appear not to be asymptotically normal. This is expected for the Firpo-Zhang estimator because Zhang (2018) showed that this estimator is not asymptotically normal in the extreme case. E.3 Dependency on k We implement simulations to investigate how the choice of the tuning parameter k = nτ n affects the MSE and the coverage of the extrapolation based extremal QTE estimators. We also present the result of the Firpo-Zhang estimator for MSE and Zhang's b out of n bootstrap for coverage as comparison. The considered models H 1 , H 2 and H 3 are the same as in Section 4. Figure 6 shows the simulation results about MSE. The MSE of the Firpo-Zhang estimator is a line because it does not depend on k. From this figure, we can see that the value of k has a big influence on the MSE of our extremal QTE estimator and the quantile extrapolation method with the causal Pickands estimator. In particular, a clear bias-variance trade-off with respect to k is shown in the plots related to H 1 and H 3 . We also note that for the more heavy-tailed models We can see that there is always a range of k where our proposed confidence interval (12) has good coverage. The particular range, however, depends on the respective model. We also note that our method works well in terms of coverage for a wide range of k for models H 2 . The above observations generally agree with the observation from classical quantile extrapolation setting, see e.g. de Haan and Ferreira (2007). G Proofs We mention again that unless otherwise stated, τ n denotes the intermediate quantile which satisfies τ n → 0 and k := nτ n → ∞. We will use notations k and τ n interchangeably for convenience. G.1 Proof of Lemma 1 To prove Lemma 1, we first introduce Theorem 5, Lemma 5, Lemma 6, Lemma 7 and Lemma 8. Theorem 5 is a special case of Theorem 3.1 in Zhang (2018), so we omit its proof. Theorem 5. Suppose that Assumptions 1, 2, and 7 hold, and assume that nτ n → ∞ and τ n → 0. Let and consider the random vector Then for j = 0, 1, and In particular, if Assumption 8 holds, then where N is bivariate Gaussian vector with mean zero and covariance matrix with H 1 , H 0 and H 10 defined as in Assumption 8. Lemma 5 shows that for a CDF F j with positive extreme value index, the normalization sequence λ j,n can be replaced by a simpler expression. Lemma 5. For j = 0, 1, suppose Assumption 2 is met and F j has an extreme value index γ j > 0. Then Proof of Lemma 5. By Assumption 2, F j satisfies the max-domain of attraction condition with a positive extreme value index γ j and its density f j is monotone in the upper tail. Therefore, the von Mises condition lim t→∞ tf j (t) 1 − F j (t) = 1 γ j holds by Theorem 2.7.1 in de Haan (1970). So we have where the second last equality is obtained by setting t = q j (1 − τ n ). Lemma 6 is a classical result in causal inference literature, and we omit its proof. Lemma 6. Let g be a measurable function such that E [\|g(Y (1))\|] and E [\|g(Y (0))\|] are finite. Suppose (Y (1), Y (0)) \|= D \| X and there exists c > 0 such that c < Π(X) < 1 − c almost surely. Then we have Lemma 7 gives the convergence rate of the estimated propensity score by using the sieve method. Lemma 7. Suppose Assumptions 1 and 7 are met. Then we have In particular, this implies Proof of Lemma 7. By Assumption 7 iv) ζ(hn) 2 hn √ n → 0, we have ζ(hn) 4 n → 0 since h n → ∞. Therefore, we can apply Lemma 1 and 2 in Hirano et al. (2003) to obtain sup x∈Supp(X) The condition ζ(hn) 2 hn In particular, we have sup x∈Supp(X) \| Π(x) − Π(x)\| = o p (1). The second part of the lemma then follows from the assumption that Π(x) is continuous and bounded away from zero, which allows us to apply the continuous mapping theorem (see Theorem 7.25 in Kosorok (2007)). Lemma 8 shows that the following two terms converge to zero in probability. This lemma will be used many times in the remaining proofs, so we prove it here. Lemma 8. Suppose Assumptions 1, 2 7 and 8 hold. Then Proof of Lemma 8. We show the first part of the lemma, and the second part follows analogously. For simplicity of notation, we denote t n = 1 − τ n . So by the triangle inequality. Now we show that both terms (27) and (28) converge to zero in probability, which then proves the original claim. For the first term (27), we need the subgradient condition for q 1 (defined by (6)). Specifically, since is convex, a necessary condition for q 1 (t n ) = arg min q∈R L n (q) is that 0 ∈ ∂L n ( q 1 (t n )) where ∂L n (q) denotes the set of subgradients of L n at q (for details see e.g. Bubeck (2015)). Because Y i is a continuous random variable for i ∈ {1, . . . , n}, there exists at most one Y i = q 1 (t n ) almost surely. In the first case where Y i = q 1 (t n ) for all i ∈ {1, . . . , n}, the subgradient condition implies In the second case where there exists some i 0 ∈ {1, . . . , n} such that Y i 0 = q 1 (t n ), we have where [−t n , 1−t n ] is an interval and the set addition is understood elementwise. The subgradient condition then implies that there exists some t ∈ [−t n , 1 − t n ] such that Combining the above two cases, we have that almost surely By Assumption 1, we have that sup x 1 Π(x) < 1 c . So by Lemma 7 we have Because nτ n → ∞, we have For the second term (28), the proof of Theorem 3.1 in Zhang (2018) showed that the term converges in distribution to a normal random variable (in particular, Zhang (2018) proved the corresponding result for lower quantiles, but the same holds for upper quantiles). Hence, we have Now we give the proof of Lemma 1. Proof of Lemma 1. We show the claim for j = 1, and the case of j = 0 can be proved analogously. First, we expand γ H 1 (defined by (7)) as In the following, we will show that G 1 n P −→ γ 1 and that G 2 n , G 3 n , G 4 n converge to zero in probability, which then proves the original claim. Now we prove that G 1 n P −→ γ 1 . This part of the proof is similar to the proof of Theorem 3.2.2 in de Haan and Ferreira (2007), which shows the consistency of the classical Hill estimator. First, we have 1))), so its CDF is 1 − 1/z for z ≥ 1, which then implies that log(Z i (1)) has a standard exponential distribution. Let U 1 = (1/(1 − F 1 )) ← be the tail function of Y i (1). Then we have that Y i (1) = U 1 (Z i (1)) and 1 Y i (1)>q 1 (1−τn) = 1 Z i (1)>τ −1 n almost surely. Since q 1 (1 − τ n ) = U 1 (τ −1 n ), we have that almost surely By Assumption 2 iii), F 1 satisfies the max-domain of attraction condition with extreme value index γ 1 > 0, so by Theorem 1.1.6 and Corollary 1.2.10 in de Haan and Ferreira (2007), we have that for all x > 0, Then, by the statement 5 of the Proposition B.1.9 in de Haan and Ferreira (2007), we have that for any ε, ε > 0 such that ε < 1, ε < γ 1 , there exists some t 0 such that for x ≥ 1, t ≥ t 0 , which is equivalent to log(1 − ε) + (γ 1 − ε ) log(x) < log(U 1 (tx)) − log(U 1 (t)) < log(1 + ε) + (γ 1 + ε ) log(x). Multiplying by 1 k D i Π(X i ) on both sides of the above inequality and summing up all i ∈ {1, . . . , n} with Z i (1) > τ −1 n gives us that almost surely, G 1 n lies in the interval [a, b] with Since ε and ε can be arbitrarily small, to prove G 1 n P −→ γ 1 , it is enough to show where at the second last equality we used Lemma 6, and Thus, by the weak law for triangular array (see Theorem 2.2.4 in Durrett (2013)), we have For (ii), let b n = k and S n = n i=1 log(Z i (1)τ n ) D i Π(X i ) 1 Z i (1)>τ −1 n . By the fact that log(Z i (1)) has a standard exponential distribution, we have Similarly, we have Thus, by the weak law for triangular array, we have This concludes that G 1 n P −→ γ 1 . Now we prove that G 2 n , G 3 n , G 4 n converge to zero in probability. For G 2 n , let ∆ n := log( q 1 (1 − τ n )) − log(q 1 (1 − τ n )) = log Given the previous results (30) and the fact that 1 Y i (1)>q 1 (1−τn) = 1 Z i (1)>τ −1 n almost surely, it is sufficient to show that ∆ n = o p (1). Consider where λ 1,n = n τn f 1 (q 1 (1 − τ n )) is defined in Theorem 5. By Lemma 5, we have √ kq 1 (1 − τ n ) −1 γ 1 λ 1,n → 1, and Theorem 5 implies that and consequently By the continuous mapping theorem, The last equality follows from Lemma 8 and the result (32) that ∆ n = o p (1). Since G 3 n ≥ 0, we have G 3 n P −→ 0. For G 4 n , we have that We have shown that G 1 n P −→ γ 1 and G 2 n = o p (1), so G 4 n = o p (1) by Lemma 7. G.2 Proof of Theorem 1 To prove Theorem 1, we first introduce Lemma 9, which shows that the term E S p i,j,n is of order O(τ n ). Lemma 9. For defined in Theorem 1 and for all p ∈ N, we have Proof of Lemma 9. We have already seen in the proof of Lemma 1 that log 1 1−F j (Y i (j)) follows a standard exponential distribution. Therefore, for p ∈ N, we have Thus the first claim follows. Note that S i,j,n ≥ 0 for positive γ 1 , thus the second claim follows from the Markov inequality. Now we prove Theorem 1. Proof of Theorem 1. We show the claim for j = 1, and the case j = 0 can be proved analogously. As in the proof of Lemma 1, we expend In the following, we will show that which then implies the original claim that For G 1 n , we proceed similarly as in the proof of Theorem 3.2.5 in de Haan and Ferreira (2007). As shown in the proof of Lemma 1, we have that almost surely, where Z i (1) = 1/(1 − F 1 (Y i (1))) and U 1 = (1/(1 − F 1 )) ← . By Assumption 3, we have for all with γ 1 > 0, ρ 1 < 0 and lim t→∞ A 1 (t) = 0. Equivalently, we have By the proof of Theorem 3.2.5 in de Haan and Ferreira (2007), there exists a function A such that lim t→∞ A(t)/A 1 (t) = 1 and for any > 0, there exists t 0 > 0 such that for all t ≥ t 0 , x ≥ 1, For large enough n and for i ∈ {1, . . . , n} such that Z i (1) > n/k, we can set t = n/k and x = Z i (1) · k/n. Multiplying by √ k D i kΠ(X i ) on both sides of the above inequality and summing up all i ∈ {1, . . . , n} with Z i (1) > n/k gives us that almost surely, where G 1,1 n := Let 0 < < −ρ 1 . We first show that √ kG 1,3 n converge in probability to some constant. Since √ kA 1 n k → λ 1 by assumption and A n k /A 1 n k → 1, it is enough to show that (1)>n/k . By the fact that Z i (1) has probability density function 1/z 2 on z ≥ 1, we can calculate Thus, by the weak law for triangular array, we have Because can be arbitrarily close to zero, by inequality (33), we have that almost surely, Similarly, by using the weak law for triangular array, one can obtain that Hence, we conclude that almost surely, For G 2 n , similarly as in the proof of Lemma 1, we have where λ 1,n and φ i,1,n are defined as in Theorem 5, and we applied Theorem 5 to obtain the last equality. By applying the delta method, we have Combining with result (30), we obtain Note that by Theorem 5, we have √ kG 2 n = O p (1). For G 3 n , similarly as in the proof of Lemma 1, we have Theorem 5 and the result (35) then imply that √ k(log(q 1 (1 − τ n )) − log( q 1 (1 − τ n ))) = O p (1). For G 4 n , we expand For the second term, we have by that fact that √ kG 2 n = O p (1) (see the comment below (36)) and Lemma 7. Hence, we obtain where for the second last equality we used that which can be shown by using a similar argument as on page 43. Thus, we have In order to derive the influence function which arises from using the estimated propensity score, we follow similar steps as in the proof of Theorem 3.1 of Zhang (2018). First, we rewrite √ kG 4 n = G 4,1 n − G 4,2 n + o p (1) with For G 4,1 n , note that S i,1,n ≥ 0, so we have where in the second inequality we used Assumption 1 iii.) and D i ≤ 1, and in the second last equality we used Lemma 7, result (29) and n i=1 S i,1,n = O p (k) which can be obtained by the Markov inequality and Lemma 9. Thus √ kG 4 n = −G 4,2 n + o p (1). For G 4,2 n , we expand G 4,2 n = G 4,3 n + G 4,4 n where and F X denotes the CDF of X. First, we show G 4,4 n = o p (1). For this we consider π n := arg min and the pseudo true propensity score Π n (x) = L(H hn (x) T π n ), where H hn is the vector consisting of h n sieve basis functions and L is the sigmoid function (see Section B for more details). We rewrite G 4,4 n = G 4,5 n + G 4,6 n with and we show that both terms converge to 0 in probability. For G 4,6 n , note that and thus E [G 4,6 n ] = 0. We also have where for the last equality we applied Lemma 9 which gives E S 2 i,1,n = O(τ n ) and the Lemma 1 of Hirano et al. (2003) The convergence to 0 can be obtained because Assumption n , we use the Taylor expansion to get G 4,5 n = G 4,5,1 n ( π n − π n ) + 1 2 ( π n − π n ) T (G 4,5,2 n − G 4,5,3 n )( π n − π n ) with G 4,5,1 where π n is random, lies on the line between π n and π n , and depends on X i resp. x. For G 4,5,1 n , we have where the first inequality holds because the summands of G 4,5,1 n are i.i.d. and with mean 0, the second inequality holds because \|L \| < 1, 1/Π(X i ) 4 < 1/c 4 and ζ(h n ) = sup x H hn (x) , and for the last equality we used Lemma 9. Thus E [ G 4,5,1 n ] ≤ (E [ G 4,5,1 Note that the summands of G 4,5,2 n and G 4,5,3 n are no longer independent because of π n . Therefore, for G 4,5,2 n and G 4,5,3 n , we apply the triangle inequality to obtain where we used similar arguments as for G 4,5,1 n and the fact that \|L \| < 1. By the Markov inequality, we have G 4,5,1 n = O p (ζ(h n )) and G 4,5,2 n = G 4,5,3 Under the Assumption 7, we have by the Lemma 2 in Hirano et al. (2003) that π n − π n = O p ( h n /n). Hence, by the Cauchy-Schwarz inequality, we have \|G 4,5 n \| ≤ G 4,5,1 n π n − π n + 1 2 π n − π n 2 ( G 4,5,2 n + G 4,5,3 By Assumption 7 iv) 1 √ n ζ(h n ) 2 h n → 0, we have ζ(h n ) hn n → 0 and √ kζ(h n ) 2 hn n → 0, thus G 4,5 n = o p (1), which concludes that G 4,4 n = o p (1). At this point, we have √ kG 4 n = −G 4,3 n + o p (1). For G 4,3 n , we proceed in a similar manner as for G 4,4 n . First, we decompose G 4,3 n = G 4,7 n + G 4,8 n with For G 4,8 n , we have where the second last equality holds because sup where π n lies on the line between π n and π n . Since π n is the solution of the optimization problem (16), the first order condition can be derived by differentiation of the objective function. Extending and applying the mean value theorem for Π(X i ) − Π n (X i ) leads to We define which allows us to write Now we show that both terms G 4,7,1 n and G 4,7,2 n are o p (1). For G 4,7,1 n , using the mean value theorem for L (H hn (x) T π n ) − L (H hn (x) T π n ) and the fact that \|L \| < 1, we have where in the last equality we used π n − π n = O p ( h n /n) and E [S i,1,n ] = O(τ n ). In addition, Hirano et al. (2003) showed in the proof of their Theorem 1 that Σ −1 n = O p (1) and E [ V n 2 ] = O(ζ(h n ) 2 ), and the last result implies that V n = O p (ζ(h n )) by the Markov inequality. Therefore, by the submultiplicativity of Frobenius norm we have where the last equality is implied by Assumption 7 iv) τnζ(hn) 10 hn n → 0. For G 4,7,2 n , we rewrite G 4,7,2 where the last equality is implied by Assumption 7 iv) τnζ(hn) 10 hn Now we show that both terms G 4,7,3 n and G 4,7,4 n are o p (1). For G 4,7,4 n , from previous proofs we know that Ψ hn = O p ( √ τ n ζ(h n )). In addition, Hirano et al. (2003) showed in the proof of their Theorem 1 that Σ −1 n = O(1). Together with the fact that \|Π n \| < 1, we have sup x δ hn (x) = O p ( √ τ n ζ(h n ) 2 ). Since Π is bounded away from 0 and 1 by Assumption 1, using the mean value theorem and the triangular inequality give us where the last equality follows from Assumption 7 iv) nτ n ζ(h n ) 6 h −s/r n → 0. For G 4,7,3 n , first note that we can view √ τ n δ hn (x) as the least squares approximation of √ τ n δ 0 (x) using the approximation functions L (H hn (x) T π n )H hn (x). By Assumption 4 i) that E [S i,1,n \|x] is t-times continuously differentiable with all derivatives bounded by N n on Supp(X). Thus, similarly as the Lemma 1 in Hirano et al. (2003) which follows from the Theorem 8 on page 90 in Lorentz (1966), it holds that sup x∈Supp(X) Hence, by the triangular inequality we have where the last equality follows from Assumption 4 ii). Therefore, we have and consequently Combining equations (34), (36), (37) and (38), we conclude that (ψ i,1,n + γ 1 φ i,1,n ) + o p (1). G.3 Proof of Theorem 2 Proof of Theorem 2. Let V i,n := (ψ i,1,n , ψ i,0,n , φ i,1,n , φ i,0,n ) T and let Σ n be its covariance matrix. For j = 0, 1, we write ψ i,j,n = ν i,j,n + η i,j,n with We will apply the multidimensional Lindeberg CLT to prove the claim. We first show that the expectation of V i,n is a zero vector. For ψ i,1,n , we have where the second equality follows from Lemma 6 and the last equality holds as E [S i,1,n ] = γ 1 τ n which can be seen from the proof of Lemma 9. Thus we have E [ψ i,1,n ] = 0. Analogously, one can show that E [ψ i,0,n ] = E [φ i,1,n ] = E [φ i,0,n ] = 0. Now we show that the covariance matrix Σ n converge to Σ via showing entry-wise convergence. We first compute where the second last equality is obtained by applying the law of iterated expectation and the last equality follows from Lemma 9. Similarly, we have Similar inequalities hold for all combinations of S i,1,n , E [S i,1,n \| X i ], T i,1,n and E [T i,1,n \| X i ]. G.4 Proof of Lemma 2 Proof of Lemma 2. The proof is similar to the proof of Theorem 4.3.8 in de Haan and Ferreira (2007). Denote d n := τ n /p n = k/(np n ), we have log(d n )/ √ k → 0 by the assumption that log(np n ) = o( √ k). Using q j (1 − τ ) = U j (1/τ ), we can expand By using the same arguments as in the proof of Lemma 1, we have By applying the Theorem 2.3.9 in de Haan and Ferreira (2007), we have which then implies Since for all s ∈ [1, d n ], we have By Theorem 2 and the fact that Combing the equality (39) and the fact that Combining the above results, we conclude that √ k log(d n ) G.5 Proof of Theorem 3 Proof of Theorem 3. Let d n := τ n /p n . First, we expand By Lemma 2 and Theorem 2, we have that for j = 0, 1, By applying the continuous mapping theorem and Theorem 2, we have where the notations are defined as in the description of the theorem. G.6 Proof of Theorem 4 We first introduce Lemma 10 which shows that under Assumption 6, the covariance matrix Σ simplifies to Σ. In particular, the components G 10 , J 10 , J 01 and H 10 become zero, which implies that the extrapolated quantiles Q H 1 (1 − p n ) and Q H 0 (1 − p n ) are asymptotically independent. Proof of Lemma 10. It is sufficient to show that the entries of Σ and Σ are equal. For H 1 , H 0 , G 1 , G 0 , J 1 and J 0 , the equalities are direct consequences of Assumption 6 and Assumption 1 that Π(x) is bounded away from 0 and 1. For other terms the equalities can be proved by using the Cauchy-Schwarz inequality. Now we give the proof of Theorem 4. Proof of Theorem 4. Recall that σ with H 1 , H 0 , G 1 , G 0 , J 1 , J 0 are defined in equation (25). By Lemma 1, we have γ H j P −→ γ j for j = 0, 1. By Lemma 2 and Assumption 5, we have κ P −→ κ. By Lemma 10, only the covariance terms H 1 , H 0 , G 1 , G 0 , J 1 and J 0 in Σ are nonzero. Therefore, to prove the consistency of the estimated variance σ 2 , it is suffice to show that for j = 0, 1, where G 1 , G 0 , J 1 , J 0 are defined as in Assumption 9 and H 1 , H 0 as in Assumption 8. We only show the result for j = 1, the case of j = 0 can be shown analogously. We proceed in a similar manner as in the proof of Lemma 1. For (i), we expand H 1 = H n,1 1 + H n,2 1 + H n,3 1 with H n,1 We will show that H n,1 since 1 nτn → 0 and 1 τn E P (Y i (1)>q 1 (1−τn)\|X i ) Π(X i ) → H 1 < ∞, thus H n,1 1 P −→ H 1 . We have shown that G n,1 1 P −→ G 1 and G n,2 1 = o p (1), so G n,4 1 = o p (1) follows from Lemma 7. Combining all the previous results we can conclude that G 1 P −→ G 1 . For (iii), we expand J 1 = J n,1 1 + J n,2 1 + J n,3 1 + J n,4 1 with J n,1 where ∆ n = log(q 1 (1 − τ n )) − log( q 1 (1 − τ n )). The next step is to show that J n,1 1 P −→ J 1 and that J n,2 1 , J n,3 1 , J n,4 1 converge to zero in probability, which is enough to prove J 1 P −→ J 1 . The remaining proof is similar to the one for G 1 , so we omit it. G.7 Proof of Lemma 3 Proof of Lemma 3. It is sufficient to show that Σ − Σ is a positive semi-definite matrix. Note that Σ − Σ can be written as the limit Therefore, ∆Σ n is of rank 1 and has at most one non-zero eigenvalue. In addition, since all entries on the diagonal of ∆Σ n are non-negative, we have trace(∆Σ n ) ≥ 0, which implies that ∆Σ n is positive semi-definite. Linearity and monotonicity of the expectation then yield that E [∆Σ n ] /τ n is positive semi-definite, which implies the positive semi-definiteness of the limit Σ − Σ.	2021-10-14T01:15:35.705Z	2021-10-13T00:00:00.000	{ "year": 2021, "sha1": "797084db2149fdf1963891a43933e359ccbbfefc", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "797084db2149fdf1963891a43933e359ccbbfefc", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics" ] }
121194777	pes2o/s2orc	v3-fos-license	High-coercivity ultralight transparent magnets Magnetic silica-aerogel composites have been synthesized by dispersing hard magnetic Nd2Fe14B particles in a sol during a fast sol-gel process and subsequently supercritically drying the resulting gels. The composites are found to retain most of the outstanding properties of their constituents: the large coercivity and moderate remanence of the magnetic powders and the transparency and low density of silica aerogels. Moreover, aerogels synthesized in the presence of a magnetic field exhibit the alignment of the particles, forming needle-like structures along the direction of the applied magnetic field, which results in optical and magnetic anisotropies. Due to their unique combination of properties, these types of materials may be appealing for magneto-optics and magnetic actuator applications. Aerogels are highly porous solid materials with unusually low densities and high specific surface areas.They are usually prepared by the supercritical drying of highly crosslinked organic or inorganic gels. 1 Gels derived from the condensation and hydrolysis of metal alkoxide precursors allow the production of a large variety of aerogels, from which silica aerogels have been extensively investigated.Silica aerogels are transparent materials with low-refractive index and low dielectric constant.8][9] Composite aerogels mainly consist in homogeneous dispersions of particles hosted in an aerogel matrix.Aerogels with particles in the nanometer range can be obtained during the gel synthesis from precursors with alkoxide functionalities, since the aerogel nanometric pores limit the particle growth. 10However, composite aerogels can also be prepared from nonbonding solid particles added to the sol during the gelation. 113][14][15][16][17] Most of these magnetic aerogel composites are formed from soft magnetic particles homogeneously dispersed in the aerogel matrix.Consequently, many of the studied systems exhibit low remanence and coercivity at room temperature, thus limiting their possible applications. 16Moreover, it has been recently shown that anisotropic distributions of magnetic particles in aerogel matrices can be prepared by using a magnetic field during the gelation process. 17Nevertheless, even in aligned aerogels, the room temperature magnetic properties remain poor.Taking into account that magnetic aerogels combine the intrinsic properties of the embedded magnetic particles with the superior properties ͑e.g., low density, transparency, and electrically insulator͒ of aerogels, it would be interesting for certain applications ͑e.g., magneto-optical modulation or magneto-optical displays, microwave absorbers, isolators, or modulators͒ 18 to have composites of high remanence and high coercivity.In this letter we demonstrate the feasibility to fabricate homogeneous and aligned magnetic aerogel composites containing Nd 2 Fe 14 B particles, which result in ultralight transparent magnetic materials with large coercivities and moderate remanence. Silica composite aerogels were obtained by supercritical drying of gels prepared by adding different amounts of commercial Nd 2 Fe 14 B particles in a viscous sol.For the gel synthesis, 6.7 ml of a sol consisting of a water solution of ammonia plus a mixture of methanol and tetramethoxysilane ͑TMOS͒ was poured into 1 cm diameter polystyrene test tubes.The molar ratios of the reactants were TMOS: methanol:water:ammoniaϭ1:12.25:4:6.5ϫ10Ϫ2 .This chemical route yields the gelation of the mixture in about 3-4 min.After the addition of the water and the catalyst (NH 3 ) the sol was softly stirred and NdFeB particles, with sizes ⌽Ͻ5 m, were added to the viscous solution.During the gelation process, to obtain a gel with a homogeneous dispersion of particles, the test tubes were closed and repeatedly reversed every 4 -5 s.Following this procedure three composites A, B, and C containing respectively 50, 300, and 500 mg of NdFeB powders diluted in 6.7 ml of sol were prepared in order to obtain composites of different magnetic a Electronic mail: mgich@icmab.esphase concentrations.Furthermore, aiming to investigate the effects of in-field synthesis, additional samples (A f , B f , and C f ) of the same compositions and prepared following the same procedure were placed in an 8 kOe homogeneous magnetic field when the viscosity was so high that no particle sedimentation was observed.The direction of the magnetic field, was orthogonal to the test tube axis and the samples were kept under field for 5 h.The gels were then removed from their tubes and aged in a methanol solution at room temperature for 10 days.The central and more homogeneous parts of gel monoliths were selected and supercritically dried.In this process the gel solvent ͑methanol͒ was exchanged by liquid CO 2 at 100 bar and 23 °C.After the solvent exchange, pressure and temperature were set over the CO 2 critical point ͑72.9 bar and 31 °C), where the liquid CO 2 filling the pores completely transformed to gas.Finally, the pressure was decreased keeping the gel structure with CO 2 gas inside the pores that after opening the high pressure plant were filled with air.A supercritical drying with methanol was attempted at 239.45 °C but resulted in a deterioration of the composite magnetic properties due to the thermallyinduced structural changes in the Nd 2 Fe 14 B particles. The obtained NdFeB composite aerogels are cylindrical monoliths.The appearance of the samples depends on the content of magnetic particles and on whether they have been synthesized in-field or without field.It is observed that in the aerogels synthesized in the presence of a magnetic field the magnetic particles are aligned forming needle-like structures along the direction of the magnetic field applied during the gelation process.This indicates that when the gels were placed under the magnetic field, the particles were still able to slightly move and consequently align following the field lines.In contrast, the aerogels prepared in absence of a magnetic field keep the relatively homogeneous suspension of magnetic particles from the gelation.This can be clearly observed in Fig. 1 Moreover, the transparency of the composites decreases when the NdFeB content increases.In fact, only the monoliths of samples A and A f , having the lowest powder content, are found to be transparent in all the directions across the cylinder axis.However, for the in-field aerogels the transparency increases along the particle alignment direction and decreases in the direction perpendicular to it ͑see Fig. 2͒.Samples B f and C f exhibit a transparency similar to that of A when the light is applied parallel to the orientation direction, although the density of particles is 6 and 10 times larger, respectively.However, when the light is perpendicular to the alignment direction samples B f and C f appear opaque.The anisotropic transparency presented by the in-field composites could make them suitable for certain applications in magnetooptics ͑e.g., based on dichroism or birefringence͒, which have been largely investigated in other magnetically anisotropic systems. 19s can be seen in Table I, the in-field composites exhibit systematically a larger density than the homogenous ones.This effect can be explained by the particles tendency during the gelation to accumulate close to the center of the gel where the magnetic field is higher. Surface area measurements by N 2 isotherm adsorption and Brunnauer-Emmet-Teller ͑BET͒ analysis were performed ͑Table I͒, and they show typical aerogel surface area values, ranging from 490 to 810 m 2 /g.The samples with higher powder contents have lower surface areas, as would be expected.The lower surface areas measured for in-field synthesised composites as compared to the samples produced without field can be due to higher presence of pores bigger than 100 nm, not detected by the BET technique.These are probably induced during the synthesis by the displacement of the magnetic particles under the action of the field. Room temperature hysteresis loops have been performed for all the samples, in fields up to 120 kOe, in a vibrating sample magnetometer.The in-field composites were measured with the field applied parallel and perpendicular to the synthesis direction.All composites display the typical features of a hard magnetic phase, i.e., large coercivity, H C , and moderate squareness, M R /M S , ͑Fig.3͒, thus indicating that the hard magnetic properties of the NdFeB powders have not been altered during the composite aerogel synthesis.All the studied aerogels exhibit coercivities in the range of H C ϳ13 kOe, i.e., well within the range of coercivities characteristic of melt spun Nd 2 Fe 14 B 20 and squareness values just above M R /M S ϭ0.5 that are typical of isotropic magnets with exchange interactions. 21The effect of particle alignment is evidenced in the difference of the values of M R /M S be- tween the composites synthesized in-field or without field. For the in-field synthesized samples, when the measuring field was applied at 0°with respect to the powders alignment direction, M R /M S ϭ0.58 was obtained, whereas when the measurements were carried out at 90°the squareness decreased to 0.53-0.54͑see inset Fig. 3͒.In the case of the samples synthesized without field one obtains values of 0.54 -0.55.The small difference between the parallel and perpendicular squareness evidences a certain degree of magnetic alignment.This shows that, besides physical orientation of the particle powders along the field lines, some magnetic anisotropy can be induced in the samples.Note that since the particles of several micron in size are multidomain and composed of several randomly oriented crystallites a full magnetic orientation is impossible in these samples.However, full magnetic and physical alignment should be achievable for single domain particles.Moreover, the energy product, (BH) max , was estimated from the hysteresis loop, where a value of 0.06 kJ/m 3 was obtained for C f .Yet, typical BH max values for NdFeB powders are of the order of 100 kJ/m 3 , about 1700 times higher.This difference can be understood on the basis of the linear dependence of the saturation magnetization, M S , with the composite bulk density ͑see inset of Fig. 3͒ and the fact that BH max scales with M s 2 ; keeping the magnetic powders content constant and reducing the sample density by a factor of about 40 ͑from 7.8 g/cm 3 of the NdFeB powders to the 0.2 g/cm 3 of the composite C f ), can explain a lowering of the BH max by a factor of 1600, in agreement with our experimental values. Sample C f shows the highest remanence of the studied composite aerogels, which is around M R ϳ8 emu/cm 3 ͑i.e., 4M R ϳ100 G).The fields at the surface of the composite aerogel monoliths, after magnetization in a pulsed field of about 30 kOe, were measured using a gaussmeter.Samples C f and B f gave values in excess of 40 and 15 Oe, respectively.Even though the magnetic fields created by the magnetic aerogels composites can be considered somewhat small, they are sufficient to attract small stainless steel pieces ͑e.g., needles͒.Moreover, in spite of their low remanence and due to their low density, these aerogels are rather sensitive, e.g., they easily move or they give magnetomechanical response, even to very small fields. In conclusion, we have demonstrated that it is possible to fabricate homogeneous and aligned magnetic aerogel composites from Nd 2 Fe 14 B particles.The composites exhibit excellent magnetic and aerogel properties, e.g., large coercivity (H C ϳ13 kOe), moderate squareness (M R /M S ϳ0.55), low density (ϳ0.14-0.22g/cm 3 ) and transparency.Due to this combination of properties this type of material should be considered as a potential candidate for applications in the fields of magneto-optics and magnetic actuators.FIG. 3. Hysteresis loop for sample C f with the applied field parallel to the particle alignment direction.The inset shows the saturation magnetization, M S , for ͑͒ in-field samples with the field applied at 0°from the synthesis direction, ͑᭝͒ in-field samples with the field applied at 90°from the synthesis direction and (*) samples synthesized without field.Shown in the inset is the normalized magnetization for sample B f along ͑solid line͒ and perpendicular to the aligning direction ͑dashed line͒. FIG. 1 . FIG. 1. Photographs of samples B ͑a͒ and B f ͑b͒ using a 50ϫ magnification. TABLE I . Bulk density and surface area of magnetic aerogel composites synthesized in-field and without field with different amounts of Nd 2 Fe 14 B powders.	2018-11-08T01:04:59.744Z	2003-06-10T00:00:00.000	{ "year": 2003, "sha1": "79402c70e69621796acb547a9a979e0d31acef81", "oa_license": "CC0", "oa_url": "https://ddd.uab.cat/pub/artpub/2003/116252/appphylet_a2003m6v82n24p4307.pdf", "oa_status": "GREEN", "pdf_src": "Anansi", "pdf_hash": "79402c70e69621796acb547a9a979e0d31acef81", "s2fieldsofstudy": [ "Materials Science" ], "extfieldsofstudy": [ "Physics" ] }
211813047	pes2o/s2orc	v3-fos-license	The Principle and Legal Regulation for "Big Data Swindle Acquaintances" —Big Data Swindle Acquaintances refers to an behavior that network platform operator may do to partial stable consumers by analyzing and generating specific users' image after mastering users' personal information and consumption data. This behavior violates the principle of honesty and credit in the civil law, and the consumer protection law, and seriously damages the rights and interests of consumers. INTRODUCTION The Social Survey Center of China Youth Daily made an online questionnaire survey in March 2018. According to the survey results from 2008 respondents, 51.3% of them had suffered from "Big Data Swindle Acquaintances" by Internet enterprises, and 63.4% of them believed that nowadays it had been a common thing that Internet enterprises took advantage of the big data to swindle acquaintances. Not long ago on Weibo, many passengers disclosed their experience of being swindled by drivers providing online car-hailing service, opened up the prelude to crusade against such behavior of "Big Data Swindle Acquaintances" and also brought the originally unknown term "Big Data Swindle Acquaintances" to vision of the public. Gradually, some users find that they are charged at different rates from the same origin to the same destination; even when consumers are booking hotel room on the network platform of a travel service company, the same room is priced differently on different mobile phones. For those Internet operation platforms, considering that their stable consumers have not that high perception and sensitivity to the prices shown, they often take this advantage to damage consumers' rights and seek for improper benefits. At present, mobile phone, computer and network have become inseparable from people's daily life. Even without exception, all applications are suspected of trying to collecting users' personal information, including but not limited to users' mobile phone number, head portrait, common nicknames and positioning. After acquiring, analyzing and processing such information, it is not a problem for those applications to swindle acquaintances. For example, a movie ticket purchased on the movie purchase application of VIP members is even more expensive than that purchased in the theater; the same type of video software purchased via IOS system is more expensive than that purchased via Android system for more than ten Yuan; if you buy travel ticket via a travel APP, you may found that the price increases with the increase of search frequency so that you may purchase it quickly for fear of continuous price increase. When you finish it, you may found that the price is reset to the beginning data. All the signs show that the environment people live in is a transparent environment. Many common applications are attempting to define consumers so as to set patterns for consumers and further swindle them. A. Price Discrimination Theory in Economics Price discrimination is a proper term in economics. In general, it is to price differently for different people groups. For example, when those driving a globally limited luxury car come to a street stall to eat noodle, they may be charged 50 Yuan a bowl of noodle; however, if they are management personnel from a company carrying a bag of famous brand and having exquisite makeup, they may be charged 30 Yuan of noodle; and if they are students and ordinary workers, they may be charged 10 Yuan a bowl of noodle. This is they way of pricing differently for different people, and hence the price discrimination. Although it is theoretically simple and easy to operate, no or very few noodle restaurant bosses would have behavior of "price discrimination" in real life. Why? It is mainly due to the constriction posed by objective laws and market competition. On the one hand, nowadays in the operation of physical store, it is required to give clear price tagging; on the other hand, consumers are very sensitive to blatant price discrimination. Even rich people cannot accept a commodity priced far higher than the market price. B. Technical Analysis of the "Big Data Swindle Acquaintances" To try to analyze the phenomenon of "Big Data Swindle Acquaintances", the first necessary step is to make clear the characteristics of the era of big data and the era of digital economy. Science and technology are the primary productive forces. With the continuous development of science and technology, people have undoubtedly entered the digital economy era. In the G20 meeting held in Hangzhou in 2018, the paper "G20 Digital Economic Development and Cooperation Initiative" defines the digital economy in this way; namely, the digital economy refers to a series of economic activities using digitalized knowledge and information as the key production factor, taking modern information network as the important carrier and taking the effective use of information communication technology as the important force for improving efficiency and optimizing economic structure. Hence, it can be seen that the three major factors of the digital economy are: data factor, network carrier, and technology power. In the context of traditional economy, asymmetric information appears between merchants and consumers due to lacking information. However, in the context of digital economy, every consumer busies with generating data like bees. The data produced by consumers is gathered on a platform. There is information asymmetry between the platform and consumers in the context of data blowout. A large amount of data makes the network platform possible to make accurate prediction. The information asymmetry makes the blind trust of consumers, and the unceasingly optimized algorithm improves the precision of marketing. Consumers having a little online shopping experience may find that when they log in a shopping platform again after they search for a product on the platform, they may receive many similar commodities as recommended by the platform, and even the picture of product you want to buy may be shown on the homepage of the website. This is mainly the so-called precise marketing towards consumers after making big data analysis. The Internet platform can also judge users' consumption power, consumption habits and consumption preferences, and give them differentiated pricing by collecting, analyzing and processing various data information of consumers. The tens of thousands of users' profile information and data as collected by the platform when user register on and use the platform, including users' device, personal account number, nicknamed avatar, geographic location information, search history, and consumption preferences and so on are the cornerstone of "Big Data Swindle Acquaintances". Without big data, there is nothing to talk about such online price discrimination. Mr. Wang Wei, the director of the Department of Information Security of Beijing Jiaotong University, said that, every participant of online life has at least hundreds of keywords. Based on those keywords, it is available to finally portray the participant and form a vivid and live image of the participant. Generally, the big data analysis on individual user can be divided into four levels: social network analysis, text analysis (for example, the dynamic state issued by users, the words most frequently used by users, and users' comment on the service), behavior analysis (such as users' social network usage habits, usual work and rest schedules, online consumption habits, etc.), and time-space sequence analysis (such as the hotel or shopping plaza that users often locate, the location of the home set by user, the location of user's work unit, etc.). Through the four levels of analysis, all aspects of user can become relatively transparent and clear. In this sense, people in the era of big data are transparent and naked. It becomes easy for online seller to analyze individual user based on those big data and select the person that is possible to become the object of "Big Data Swindle Acquaintances". A. "Big Data Swindle Acquaintances" Is Suspected of Infringing on the Legitimate Rights and Interests of Consumers The Consumer Protection Act stipulates that "consumers have the right to know the true state of the goods they purchase or use, or the services they receive, the right to independently choose goods or services, and the right to enjoy fair trade". "Big Data Swindle Acquaintances" violates consumers' right to know, right to select independently, right to enjoy fair trade. Unlike a bottle of cola, which sells only 3 Yuan in a supermarket but 20 Yuan in a star-rated hotel, the normal market price of cola is transparent, so the premium in a star-rated hotel is disclosed. In the phenomenon of "Big Data Swindle Acquaintances", stable users of various network platforms are provided with quotation higher than other users of the same period for the same service; users can neither know the pricing principle nor enjoy the fair trade right; the platform may show different optional commodities for different users. This is caused for reason that the platform controls the visibility of products so that users lose the right to independently select product and service as demanded. In addition, "Big Data Swindle Acquaintances" is also suspected of posing price fraud. According to the "Provisions for Prohibiting Price Fraud Behavior", it can be known that fraud behavior means that operator uses false or misleading price tagging forms or price means to deceive or induce consumers or other operators to transact with him/her. For those users getting VIP qualification at cost under the induction of online taxi software and movie ticket sales platform, they should be like regular customers in real life, and able to enjoy more conveniences and benefits, but in fact, they are charged higher than ordinary users. This behavior is clearly consistent with the characteristics of price fraud and thus belongs to price fraud. B. It Is Also Suspected of Abusing Users' Information In the phenomenon of "Big Data Swindle Acquaintances", an online travel service improvement platform provides different customers with different optional flight schedules, ticket fares, room types and prices, mostly based on those users' previous historical travel data. The reason behind this is mostly because the data of users on different websites is shared by ads Union. A large amount of information is used, analyzed and intensively explored by many companies without the permission or authorization of users, forming accurate user portraits to facilitate the companies to charge higher prices to users with basic demand or low price sensitivity. "Big Data Swindle Acquaintances" also reflects that Internet enterprises have large problems in aspects of data collection, user privacy protection and data circulation. Advances in Economics, Business and Management Research, volume 94 IV. COUNTERMEASURES FOR COPING WITH "BIG DATA SWINDLE ACQUAINTANCES" A. Establishing Applicable Sound Laws and Regulations It is necessary to improve relevant laws and regulations and complement the blank part as soon as possible, and formulate strict provisions for the collection, use and protection of big data. The purpose is on the one hand to adapt to the new environment brought by the emerging big data industry, and better match with the business needs based on big data analysis, on the other hand, to provide legal support to crack down the "Big Data Swindle Acquaintances" behavior and enhance the legal awareness of online platform merchants. First, it is needed to increase the popularity of ecommerce and pricing related regulations in relevant laws such as the Price Law and the Anti-Monopoly Law, so that consumers can understand China's regulations on the operation of commercial activities and use law to protect their rights. Second, it is needed to carry out clear legal characterization and regulation on "Big Data Swindle Acquaintances" behavior, increase the cost of violating law, and stipulate that the network platform should publicize its platform service agreement and payment rules. Third, in the Consumer Protection Law, the duty of proofing on "Big Data Swindle Acquaintances" should be reversed, and in turn, the merchants should be responsible for proofing that they do not implement price discrimination. Fourth, citizens should be given the rights to be forgotten, timely notified, etc., so that citizens can more effectively protect their interests. B. Strengthening Supervision by Authorities First, it is needed to build up a big data monitoring platform to monitor the phenomenon of "Big Data Swindle Acquaintances". That is to use the data analysis function of big data to judge whether an enterprise is suspected of having "Big Data Swindle Acquaintances" behavior, and then feedback the analysis result to user. Second, it is needed to establish a national big data information development department that integrates review, supervision and governance to strengthen the management and control of big data development, improve management and control efficiency, strengthen pertinence and overall planning, shorten the problem processing cycle, and provide citizens with reporting channels. Third, it is needed to newly set up a blacklist system against the "Big Data Swindle Acquaintances" behavior online. The national network supervision department should network with the media, and circulate throughout the Internet a notice about the merchants who have "Big Data Swindle Acquaintances" behavior twice or more by reference to the action executed on the person against who a judgment or order is being executed for his/her untrustworthy behavior, and the media should announces to the public to warn the platform merchants. To reform the market supervision system, the regulatory authorities should improve the supervision methods and conduct fair investigations and discover the truth. In the context of the constant emergence of new problems, it is necessary to formulate new supervision methods and conduct unified and orderly market supervision. Further, it is needed to realize a full cooperation between the national regulatory authorities and the network management departments, to not only guide the merchants to operate in good faith, but also resolutely crack down and rectify the illegal behaviors and create a good network environment. C. Improving the Awareness of Right Protection While purchasing products and services online, consumers should improve their sensitivity to price, do not rely too much on a certain platform, but pay attention to shop around and try to compare and experience the same product or service on different platforms. At the same time, consumers should also pay attention to the price fluctuations, pay attention to safeguarding rights and interests and protecting privacy. Consumers should make good use of reporting manner to announce any case violating consumers' legal rights and interests via TV news, online media and other ways of communication. Moreover, it is also necessary to resolutely establish a sense of rights protection to put to any merchants' illegal behaviors to an end. V. CONCLUSION As Dickens written in A Tale of Two Cities, "It was the best of times, it was the worst of times; It was the age of wisdom, it was the age of foolishness; It was the epoch of belief, it was the epoch of incredulity; It was the season of light, it was the season of darkness; It was the spring of hope, it was the winter of despair; We had everything before us, we had nothing before us...". The era of big data is such an era. Along with the east wind of the scientific and technological revolution, social life has undergone earth-shaking changes. On the way of regulating the Internet business behaviors and protecting consumer rights, national regulatory authorities, Internet companies, and consumers still have a long way to go.	2019-10-03T09:11:18.176Z	2019-01-01T00:00:00.000	{ "year": 2019, "sha1": "0595d11a71b2ca3095fbb07a61ff173411f1a597", "oa_license": "CCBYNC", "oa_url": "https://doi.org/10.2991/jahp-19.2019.87", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "7c6d3bd273b8f498fa8e179c8f39a9e320d4f4df", "s2fieldsofstudy": [ "Law", "Computer Science" ], "extfieldsofstudy": [ "Business" ] }
246199029	pes2o/s2orc	v3-fos-license	CARD9 Forms an Alternative CBM Complex in Richter Syndrome Simple Summary The transformation process of chronic lymphocytic leukemia into an aggressive lymphoma, called Richter syndrome (RS), is incompletely understood, and therapeutic options are limited. Here, we report CARD9 to be expressed in a subset of RS tissue specimen and in the first and only available RS cell line, U-RT1. In U-RT1, CARD9 attaches to BCL10 and MALT1, and knockdown of CARD9 leads to a significant reduction in cell viability. We hypothesized that CARD9 plays an oncogenic role in RS through the activation of NF-κB signaling. Our findings may help to extend the current knowledge about the pathogenesis of RS and promote the development of targeted therapies for this aggressive disease. Abstract Richter syndrome (RS) is defined as the transformation of chronic lymphocytic leukemia (CLL) into an aggressive lymphoma, mostly diffuse large B-cell lymphoma (DLBCL). Despite intensive therapy, patients with RS have an unfavorable clinical outcome. The detailed pathobiology of Richter transformation still needs to be elucidated. Here, we report high mRNA and protein levels of CARD9 in the RS cell line U-RT1. Co-immunoprecipitation revealed the assembly of a CBM complex using CARD9 instead of CARD11. CARD9 is known to be an activator of NF-кB signaling in myeloid cells. U-RT1 Western blot analyses showed phosphorylation of IκB as well as IKK, indicating a constitutively active canonical NF-кB pathway. This was further supported by the significant reduction in cell viability and CYLD cleavage products after CARD9 siRNA knockdown. We also showed immunostaining for CARD9 in 53% of cases analyzed in a series of RS tissue specimens, whereas other lymphomas rarely show CARD9 expression. This is the first report on ectopic expression and function of CARD9 in an aggressive B-cell lymphoma. Our findings suggest that CARD9 may contribute to the pathogenesis of RS. Introduction Chronic lymphocytic leukemia is an indolent B-cell neoplasm that, in 5-10% of cases, will undergo malignant transformation into an aggressive B-cell lymphoma, most commonly a diffuse large B-cell lymphoma [1,2], termed Richter syndrome (RS) [3]. Diagnosis requires a histologic shift to DLBCL that must be distinguished from an "acceleration" of CLL with expansion of proliferation centers [4]. Besides the DLBCL variant, a classical Hodgkin lymphoma (cHL) variant of RS can occur, but unusual variants such as plasmablastic lymphoma and high-grade T-cell lymphoma have also been described [5,6]. RS DLBCL cells invariably express CD19, but tend to lose the expression of CD5 and CD23, leading to a morphologic and immune phenotype different from the underlying CLL [7]. Using the Hans algorithm [8], 90-95% of RS DLBCL are identified as the more aggressive activated B-cell subtype (ABC DLBCL) [1,7,9]. Despite intensive therapy, patients with RS have a poor clinical outcome, with a median survival of 1-2 years from diagnosis [10,11]. A Cell Culture Cell culture reagents were obtained from Lonza (Basel, Switzerland), and fetal calf serum from Seromed/Biochrom (Berlin, Germany). All cell lines were grown in IMDM/RPMI (4:1) supplemented with 10% fetal calf serum, glutamine, and 100 U/mL L-penicillin/ streptomycin at 37 • C and 5% CO 2 . For the production of cell pellets, cells were precipitated with pure alcohol, subsequently fixed in 5% formalin solution, and embedded in paraffin. Microarray Gene Expression Profiling Total RNA was extracted from cell lines with the RNeasy Mini-Kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. Gene expression analysis was performed using the Human Gene Expression 4 × 44 K Microarray Kit (Design ID 014850, Agilent Technologies, Santa Clara, CA, USA). A total of 200 ng RNA per sample was used and processed according to the standard protocol (Agilent Technologies, Santa Clara, CA, USA). Analyses were performed in triplicate. Signals were captured by a Microarray Scanner (G2565, Agilent Technologies), and data were analyzed with Genespring 12.1 (Agilent Technologies). Expression data of U-RT1 were compared with published data on the HBL-1, OCI-Ly3, OCI-Ly7, and SUDHL6 cell lines with the following signatures: Array Express E-MTAB-973 and GEO Accession Number GSM466597. Quantitative Real-Time PCR The mRNA expression levels were determined by real-time reverse transcriptase (RT)-PCR. β-Actin and GAPDH served as internal controls. A NanoDrop 2000 spectrophotometer (Thermo Fisher) was used to measure RNA quality and quantity. The SuperScript II system (Invitrogen, Carlsbad, CA, USA) was used to synthesize single-strand cDNA from oligo-dT-primers according to the manufacturer's instructions, and qPCR was performed with the QuantiTect SYBR Green PCR Kit (Qiagen) using a LightCycler Rotor Gene Q (Qiagen). For all target genes, exon-exon boundary primers were designed to minimize amplification of traces of DNA. Primer sequences are documented in Supplementary Table S1. The PCR reaction was performed as follows: the initial denaturation step was at 95 • C for 5 min, followed by 45 cycles at 95 • C for 10 s and 60 • C (CARD9, CARD11, MALT1, and BCL10) or 51 • C (β-Actin) for 20 s. A melting curve analysis was performed for each experiment, and PCR products were analyzed by agarose gel electrophoresis to confirm specificity. All experiments were performed in triplicate. Quantification was performed using the established ∆∆CT method. Gene Sequencing DNA was isolated from U-RT1 cells with the QiAmp DNA Mini Kit (Qiagen). Gene sequencing of MYD88 (exon 5) and CD79A (exon 5 + 6) was performed according to a standard protocol for diagnostic purposes using a Multiplex PCR Kit (Qiagen, cat. no. 206143) and MiSeq-Sequencer (Illumina, San Diego, USA, cat. no. Y-410-1003). Primer sequences are documented in Supplementary Table S1. For sequencing CARD9 (all exons) and CARD11 (all exons), a custom-made Gene Read V2 chemistry panel (Qiagen) was applied. Target enrichment, amplicon processing, and library generation were performed as suggested by the manufacturer. Clonality Testing (Heavy Chain Rearrangement) The rearrangement of the immunoglobulin heavy chain gene was determined by multiplex-PCR using framework region 1, 2, and 3 (FR1, FR2, and FR3)-specific primers as described by Dongen et al. (2003 Leukemia) and as previously described in Schmid et al. [24]. Western Blot and Co-Immunoprecipitation Protein isolation was performed in TNT++ buffer (20 mM Tris pH 8.0, 200 mM sodium chloride, 1% Triton-X100, 1 mM DTT) containing both phosphatase and protease inhibitor cocktails (Sigma-Aldrich, St. Louis, MO, USA). Lysates were snap-frozen in liquid nitrogen for 3 min and subsequently centrifuged at 14,000 rpm for 10 min at 4 • C to remove debris. For protein quantification, the BCA Protein Assay Kit (Thermo Fisher) was employed, following the manufacturer's instructions. Measurements were made in an EPOCH spectrophotometer (BioTek, Winooski, VT, USA) at a 562 nm wavelength. SDS-PAGE was performed with 10-20 µg protein using Nu-PAGE 4-12% Bis-Tris protein gels (ThermoFisher). A standard semidry method using 0.2 mm nitrocellulose Cancers 2022, 14, 531 4 of 15 membranes (GE Healthcare, Chicago, IL, USA) was applied for protein transfer. Membrane blocking was done with 5% milk powder in TBS Tween 20. Primary antibodies were incubated overnight at 4 • C. After multiple washing steps and repeat blocking with 5% milk powder in TBS Tween 20, membranes were incubated for 1 h at room temperature with horseradish-peroxidase-conjugated secondary antibody. Detection was performed using the ECL substrate WesterSure (LI-COR) on a Western Blot Scanner (LI-COR Biosciences, Lincoln, OR, USA). Evaluation was done with C-DiGit Blot Scanner Software, and quantification with ImageJ (National Institutes of Health, Bethesda, MD, USA). For co-immunoprecipitation assays, 0.5-1 mg whole protein extract was mixed with 1 mg goat anti-BCL10 antibody (C17, Santa Cruz); rabbit anti-p65 antibody (AB_2793231, Active Motif, Waterloo, Belgium) served as control. After overnight incubation at 4 • C on a rotator, 10 µg protein G (Protein G Sepharose 4 Fast Flow, GE Healthcare) dissolved in 100 µL TNT++ buffer was added, and incubation was continued at 4 • C for 1 h. Immunoprecipitates were washed twice with TNT++ buffer and twice with TBS buffer before separation by SDS-PAGE and Western blot analysis as described above. Inhibition with BAY11-7082 and siRNA Knockdown To inhibit the NF-κB pathway, cells were treated for 24 h with different concentrations of the inhibitor compound BAY11-7082 (Selleckchem, Houston, TX, USA) diluted in DMSO (Sigma-Aldrich) before an MTT cell proliferation assay was performed. As a control, cells were incubated with equivalent amounts of DMSO omitting BAY11-7082. Measurements were performed in technical quintuplicates, and each test series was repeated three times. Transfection of cell lines with siRNA was performed using the Amaxa Cell Line Nucleofector Kit V (Lonza) with program X001 according to the manufacturer's protocol. For CARD9 siRNA knockdown, a mixture of siRNA oligonucleotides (Qiagen Flexitube Gene Solution) was used. The siRNA sequences are shown in Supplementary Table S1. Then, 24 h after siRNA transfection, cell viability was measured using an MTT proliferation assay. Statistics For statistical analysis, the Welch two-sample t-test was used. A p-value ≤ 0.05% was considered significant. EC50 estimates for Bay11-7082 treatment were obtained from Emax models calculated using the stan_emax function from the R package rstanemax. In MTT experiments, the paired t-test was used. Sample values were normalized to the individual mean of the control group, to allow a comparison across all samples at the same time. All analyses were performed using R version 4.0.0 (The R Foundation, Vienna, Austria). Lymphoma FFPE Tissue Bank We analyzed FFPE samples from 15 patients with RS, 26 patients with cHL, 10 patients with de novo DLBLC, 25 patients with CLL, and 17 patients with aggressive lymphoma transformed from indolent lymphoma. All work was performed in conformity with the Declaration of Helsinki, and the study was approved by the local ethics committee (Ulm Ethics Committee, vote 371/17). The patients gave written informed consent for the scientific use of their tumor samples. U-RT1 Shows High Levels of CARD9 Expression We carried out microarray expression profiling to identify genes differentially expressed in the still-unique RS cell line U-RT1 and de novo DLBCL cell lines. This revealed a high level of CARD9 mRNA in U-RT1 compared with the ABC DLBCL cell lines HBL-1 and OCI-Ly3, as well as with the GCB DLBCL cell lines OCI-Ly7 and SUDHL4. CARD9 is known to be an important activator of NF-кB signaling in myeloid cells, forming an alternative CBM complex with MALT1 and BCL10. CARD11, however, was only weakly expressed in U-RT1 ( Figure 1A). RT-qPCR analyses confirmed these results, also revealing higher mRNA expression of the CBM complex proteins MALT1 and BCL10 and a significantly lower CARD11 expression in U-RT1 compared with the de novo DLBCL cell lines ( Figure 1B). On the protein level, Western blotting showed CARD9 exclusively in U-RT1, whereas all the de novo DLBCL cell lines and the cHL cell line L428 lacked the CARD9 protein ( Figure 1C). Immunoblotting also confirmed high levels of MALT1 in U-RT1. BCL10 protein levels in U-RT1 were comparable to the de novo DLBCL cell lines ( Figure 1D). Lymphoma FFPE Tissue Bank We analyzed FFPE samples from 15 patients with RS, 26 patients with cHL, 10 patients with de novo DLBLC, 25 patients with CLL, and 17 patients with aggressive lymphoma transformed from indolent lymphoma. All work was performed in conformity with the Declaration of Helsinki, and the study was approved by the local ethics committee (Ulm Ethics Committee, vote 371/17). The patients gave written informed consent for the scientific use of their tumor samples. U-RT1 Shows High Levels of CARD9 Expression We carried out microarray expression profiling to identify genes differentially expressed in the still-unique RS cell line U-RT1 and de novo DLBCL cell lines. This revealed a high level of CARD9 mRNA in U-RT1 compared with the ABC DLBCL cell lines HBL-1 and OCI-Ly3, as well as with the GCB DLBCL cell lines OCI-Ly7 and SUDHL4. CARD9 is known to be an important activator of NF-кB signaling in myeloid cells, forming an alternative CBM complex with MALT1 and BCL10. CARD11, however, was only weakly expressed in U-RT1 ( Figure 1A). RT-qPCR analyses confirmed these results, also revealing higher mRNA expression of the CBM complex proteins MALT1 and BCL10 and a significantly lower CARD11 expression in U-RT1 compared with the de novo DLBCL cell lines ( Figure 1B). On the protein level, Western blotting showed CARD9 exclusively in U-RT1, whereas all the de novo DLBCL cell lines and the cHL cell line L428 lacked the CARD9 protein ( Figure 1C). Immunoblotting also confirmed high levels of MALT1 in U-RT1. BCL10 protein levels in U-RT1 were comparable to the de novo DLBCL cell lines ( Figure 1D). Immunostaining Reveals CARD9 Expression in the U-RT1 Cell Line and in the U-RT1 Parent Lymph Node CARD9 is predominantly expressed in myeloid cells [25][26][27]. Accordingly, immunostaining of reactive lymph nodes revealed CARD9 in macrophages and dendritic cells, but not in lymphoid cells (Figure 2A,B). In contrast, the parent RS DLBCL from which U-RT1 was established contained sheets of CARD9-positive lymphoma cells ( Figure 2C). Double immunofluorescence staining of CARD9 and CD20 ( Figure 2D-F) demonstrated a colocalization of these two antigens, confirming the high prevalence of CARD9-positive neoplastic B cells. Immunostaining Reveals CARD9 Expression in the U-RT1 Cell Line and in the U-RT1 Parent Lymph Node CARD9 is predominantly expressed in myeloid cells [25][26][27]. Accordingly, immunostaining of reactive lymph nodes revealed CARD9 in macrophages and dendritic cells, but not in lymphoid cells (Figure 2A,B). In contrast, the parent RS DLBCL from which U-RT1 was established contained sheets of CARD9-positive lymphoma cells (Figure 2C). Double immunofluorescence staining of CARD9 and CD20 ( Figure 2D-F) demonstrated a colocalization of these two antigens, confirming the high prevalence of CARD9positive neoplastic B cells. We also tested the de novo DLBCL cell lines HBL-1 and SUDHL4 for CARD9 protein levels using immunostaining ( Figure 2G,H). In contrast to U-RT1 ( Figure 2I) these cell lines showed no immunoreactivity for CARD9. Taken together, these results confirmed the exceptional expression of CARD9 in U-RT1. CARD9 Is Expressed in a Subset of Richter Syndrome Cases We analyzed CARD9 expression in 15 RS tissue samples using immunohistochemical staining. In eight (53%) samples, expression of CARD9 was detected in the neoplastic We also tested the de novo DLBCL cell lines HBL-1 and SUDHL4 for CARD9 protein levels using immunostaining ( Figure 2G,H). In contrast to U-RT1 ( Figure 2I) these cell lines showed no immunoreactivity for CARD9. Taken together, these results confirmed the exceptional expression of CARD9 in U-RT1. CARD9 Is Expressed in a Subset of Richter Syndrome Cases We analyzed CARD9 expression in 15 RS tissue samples using immunohistochemical staining. In eight (53%) samples, expression of CARD9 was detected in the neoplastic CD20positive B-cell population. A bone marrow aspirate from one patient with RS contained both the aggressive lymphoma and the underlying CLL ( Figure 3A-F). The CLL component was represented by an infiltrate of slowly proliferating small lymphocytic cells, which were CD20-positive but negative for CARD9 staining. The aggressive RS component in the same biopsy showed strong CD20 expression, but also cytoplasmic staining for CARD9. Figure S2 shows another example of an RS case with CARD9 expression. In 7 of the 15 RS samples, we were able to study the clonal relationship between the RS and the underlying CLL, while 6 cases were shown to be clonally related RS. The one RS case that was not clonally related to the underlying CLL was among the CARD9-negative cases (Supplementary Table S2). In five of the tested RS samples, we additionally analyzed the corresponding CLL specimen for CARD9 expression. None of these CLL showed CARD9 positivity, even though three of the corresponding RS did. CD20-positive B-cell population. A bone marrow aspirate from one patient with RS contained both the aggressive lymphoma and the underlying CLL ( Figure 3A-F). The CLL component was represented by an infiltrate of slowly proliferating small lymphocytic cells, which were CD20-positive but negative for CARD9 staining. The aggressive RS component in the same biopsy showed strong CD20 expression, but also cytoplasmic staining for CARD9. Supplementary Figure S2 shows another example of an RS case with CARD9 expression. In 7 of the 15 RS samples, we were able to study the clonal relationship between the RS and the underlying CLL, while 6 cases were shown to be clonally related RS. The one RS case that was not clonally related to the underlying CLL was among the CARD9-negative cases (Supplementary Table S2). In five of the tested RS samples, we additionally analyzed the corresponding CLL specimen for CARD9 expression. None of these CLL showed CARD9 positivity, even though three of the corresponding RS did. We also compared several clinical parameters; i.e., age at CLL diagnosis, DLBCL subtype, and the number of CLL therapy cycles received, but did not find significant differences between CARD9-positive and -negative cases in this small cohort (Supplementary Table S2). CARD9 Is Rarely Expressed in De Novo DLBCL, in DLBCL Transformed from Lymphoma other than CLL, in cHL, or in CLL Using immunohistochemical staining, we found a focal CARD9 expression in 1 of 10 de novo DLBCLs tested. The de novo DLBCL group included five ABC DLBCLs and five GCB DLBCLs. We also tested 17 DLBCLs that had arisen from lymphomas other than CLL/SLL, and found CARD9 expression in 2 of these cases. Among the transformed DLBCL, 11 progressed from follicular lymphoma (FL), 5 from lymphoplasmacytic lymphoma, and 1 from a mantle cell lymphoma. In this group, both CARD9-positive cases were DLBCLs that had progressed from a follicular lymphoma. These results indicated that CARD9 expression is neither a common feature of DLBCL nor a general characteristic associated with the transformation of indolent lymphoma into DLBCL ( Figure 4A). Although most cases of RS are DLBCL, transformation to cHL may occur. We therefore tested 26 cases of cHL, among them 6 lymphocyte-rich and 2 lymphocyte-depleted subtypes, 8 mixed cellularity cHLs, and 10 cases of cHL with nodular sclerosis. Six cases were tumor recurrences. Only a single case of cHL, a mixed cellularity cHL, had CARD9- We also compared several clinical parameters; i.e., age at CLL diagnosis, DLBCL subtype, and the number of CLL therapy cycles received, but did not find significant differences between CARD9-positive and -negative cases in this small cohort (Supplementary Table S2). CARD9 Is Rarely Expressed in De Novo DLBCL, in DLBCL Transformed from Lymphoma Other Than CLL, in cHL, or in CLL Using immunohistochemical staining, we found a focal CARD9 expression in 1 of 10 de novo DLBCLs tested. The de novo DLBCL group included five ABC DLBCLs and five GCB DLBCLs. We also tested 17 DLBCLs that had arisen from lymphomas other than CLL/SLL, and found CARD9 expression in 2 of these cases. Among the transformed DLBCL, 11 progressed from follicular lymphoma (FL), 5 from lymphoplasmacytic lymphoma, and 1 from a mantle cell lymphoma. In this group, both CARD9-positive cases were DLBCLs that had progressed from a follicular lymphoma. These results indicated that CARD9 expression is neither a common feature of DLBCL nor a general characteristic associated with the transformation of indolent lymphoma into DLBCL ( Figure 4A). Although most cases of RS are DLBCL, transformation to cHL may occur. We therefore tested 26 cases of cHL, among them 6 lymphocyte-rich and 2 lymphocyte-depleted subtypes, 8 mixed cellularity cHLs, and 10 cases of cHL with nodular sclerosis. Six cases were tumor recurrences. Only a single case of cHL, a mixed cellularity cHL, had CARD9-positive Hodgkin Reed-Sternberg (HRS) cells. We furthermore tested 20 cases of CLL and did not find any CARD9 expression. positive Hodgkin Reed-Sternberg (HRS) cells. We furthermore tested 20 cases of CLL and did not find any CARD9 expression. Array CGH Reveals a Copy Number Gain of CARD9 Gene Locus We performed aCGH in our cell line U-RT1 [24] and reanalyzed the data to elucidate the genomic alterations that accounted for high CARD9 expression. We found a copy number gain of the 9q34 CARD9 gene locus ( Figure 4B), whereas there were no aberrations in gene loci of the other CBM proteins CARD11 (7p22), MALT1 (18q21), or BCL10 (1p22). Co-Immunoprecipitation Reveals Assembly of an Alternative CBM Complex Consisting of CARD9, BCL10, and MALT1 in U-RT1 We speculated that CARD9 might interact with BCL10 and MALT1 to form an alternative CBM complex in U-RT1 cells, and carried out BCL10 co-immunoprecipitation to test this hypothesis. BCL10-containing protein complexes were isolated from cell lysates of U-RT1 and HBL-1 by anti-BCL10 co-immunoprecipitation, and the interaction of BCL10 with CARD9 and MALT1 was determined by immunoblot analysis ( Figure 5A). MALT1 was copurified efficiently in both cell lines, but only U-RT1 showed coprecipitation of CARD9. The absence of CARD9 coprecipitation in HBL-1 was consistent with previous results of RT-qPCR analyses, whole-protein Western blots ( Figure 1C), and immunostaining of cell precipitates ( Figure 2G). Anti-p65 immunoprecipitation and subsequent immunoblotting with anti-BCL10, anti-p65, and anti-MALT1 antibodies revealed no coprecipitation of BCL10 or MALT1 in U-RT1, confirming the specificity of the anti-BCL10 immunoprecipitation. These results indicated the assembly of an alternative CBM complex in U-RT1, consisting of BCL10, MALT1, and CARD9. Array CGH Reveals a Copy Number Gain of CARD9 Gene Locus We performed aCGH in our cell line U-RT1 [24] and reanalyzed the data to elucidate the genomic alterations that accounted for high CARD9 expression. We found a copy number gain of the 9q34 CARD9 gene locus ( Figure 4B), whereas there were no aberrations in gene loci of the other CBM proteins CARD11 (7p22), MALT1 (18q21), or BCL10 (1p22). Co-Immunoprecipitation Reveals Assembly of an Alternative CBM Complex Consisting of CARD9, BCL10, and MALT1 in U-RT1 We speculated that CARD9 might interact with BCL10 and MALT1 to form an alternative CBM complex in U-RT1 cells, and carried out BCL10 co-immunoprecipitation to test this hypothesis. BCL10-containing protein complexes were isolated from cell lysates of U-RT1 and HBL-1 by anti-BCL10 co-immunoprecipitation, and the interaction of BCL10 with CARD9 and MALT1 was determined by immunoblot analysis ( Figure 5A). MALT1 was copurified efficiently in both cell lines, but only U-RT1 showed coprecipitation of CARD9. The absence of CARD9 coprecipitation in HBL-1 was consistent with previous results of RT-qPCR analyses, whole-protein Western blots ( Figure 1C), and immunostaining of cell precipitates ( Figure 2G). Anti-p65 immunoprecipitation and subsequent immunoblotting with anti-BCL10, anti-p65, and anti-MALT1 antibodies revealed no coprecipitation of BCL10 or MALT1 in U-RT1, confirming the specificity of the anti-BCL10 immunoprecipitation. These results indicated the assembly of an alternative CBM complex in U-RT1, consisting of BCL10, MALT1, and CARD9. Phosphorylated IKK and IкB Indicate NF-кB Activation in U-RT1 Activation of NF-кB generally involves phosphorylation and inactivation of IкB, which is catalyzed by phosphorylated and active IKK. Hence, the presence of phosphorylated IKK and IкB indicate an active canonical NF-кB pathway [28,29]. Immunoblotting revealed phosphorylated IкB in U-RT1, as well as in the ABC DLBCL cell lines HBL-1, OCI-Ly3, and OCI-Ly10, while phosphorylation of IкB was absent in the GCB DLBCL cell lines OCI-Ly7, SUDHL4, and SUDHL6. Immunoblotting for phosphorylated IKK largely showed the same pattern ( Figure 5B,C). Phosphorylated IKK and IкB Indicate NF-кB Activation in U-RT1 Activation of NF-кB generally involves phosphorylation and inactivation of IкB, which is catalyzed by phosphorylated and active IKK. Hence, the presence of phosphorylated IKK and IкB indicate an active canonical NF-кB pathway [28,29]. Immunoblotting revealed phosphorylated IкB in U-RT1, as well as in the ABC DLBCL cell lines HBL-1, OCI-Ly3, and OCI-Ly10, while phosphorylation of IкB was absent in the GCB DLBCL cell lines OCI-Ly7, SUDHL4, and SUDHL6. Immunoblotting for phosphorylated IKK largely showed the same pattern ( Figure 5B,C). Treatment with NF-кB Pathway Inhibitor BAY11-7082 Reduces Viability of U-RT1 Bay11-7082 is a potent inhibitor of the NF-кB pathway, and has been shown to induce apoptosis in a variety of cancer cells [30][31][32]. An MTT assay was performed after treatment Treatment with NF-кB Pathway Inhibitor BAY11-7082 Reduces Viability of U-RT1 Bay11-7082 is a potent inhibitor of the NF-кB pathway, and has been shown to induce apoptosis in a variety of cancer cells [30][31][32]. An MTT assay was performed after treatment with Bay11-7082 to investigate the effect of NF-кB pathway inhibition on U-RT1, HBL-1, and SUDHL4. While all three cell lines tested showed reduced cell viability after 24 h of treatment, U-RT1 cells were the most vulnerable. The GCB DLBCL cell line SUDHL4 showed the lowest sensitivity to NF-кB inhibition. Mean cell viability plotted against increasing inhibitor concentrations and EC50 values are illustrated in Figure 5D. Individual values and confidence limits are given in Supplementary Figure S1 Activating mutations of CARD11, MYD88, and CD79A are known to be hallmarks of ABC DLBCL, and are often crucial for NF-кB activation in these tumors [21,23]. We performed DNA sequencing to elucidate whether mutations of the aforementioned genes may contribute to the constitutively active NF-кB signaling in U-RT1. Known target sites of MYD88 (exon 5) and CD79A (exon 5 + 6) showed wild-type sequences in U-RT1. For CARD11, we performed whole-exon sequencing. None of the common activating mutations was detected. In addition, we conducted whole-exon sequencing of CARD9 and found no divergence from the wild-type sequence. CARD9 Knockdown Decreases Cell Viability of U-RT1 The next step was to determine the functional relevance of CARD9 in U-RT1 using siRNA knockdown. We found that even a slight downmodulation of CARD9 protein levels to 72% of control ( Figure 6A) led to a significant reduction in cell viability (p = 0.0018) ( Figure 6B). We also performed MALT1 substrate cylindromatosis (CYLD) Western blot analysis after CARD9 siRNA knockdown, and detected a significant reduction in C-terminal CYLD cleavage product compared with the control group (p = 0.017). Loss of CYLD cleavage products indicates reduction in MALT1 paracaspase activity [33]. with Bay11-7082 to investigate the effect of NF-кB pathway inhibition on U-RT1, HBL-1, and SUDHL4. While all three cell lines tested showed reduced cell viability after 24 h of treatment, U-RT1 cells were the most vulnerable. The GCB DLBCL cell line SUDHL4 showed the lowest sensitivity to NF-кB inhibition. Mean cell viability plotted against increasing inhibitor concentrations and EC50 values are illustrated in Figure 5D. Individual values and confidence limits are given in Supplementary Figure S1. DNA Sequencing of CARD9, CARD11, MYD88, and CD79A in U-RT1 Exhibits no Activating Mutations Activating mutations of CARD11, MYD88, and CD79A are known to be hallmarks of ABC DLBCL, and are often crucial for NF-кB activation in these tumors [21,23]. We performed DNA sequencing to elucidate whether mutations of the aforementioned genes may contribute to the constitutively active NF-кB signaling in U-RT1. Known target sites of MYD88 (exon 5) and CD79A (exon 5 + 6) showed wild-type sequences in U-RT1. For CARD11, we performed whole-exon sequencing. None of the common activating mutations was detected. In addition, we conducted whole-exon sequencing of CARD9 and found no divergence from the wild-type sequence. CARD9 Knockdown Decreases Cell Viability of U-RT1 The next step was to determine the functional relevance of CARD9 in U-RT1 using siRNA knockdown. We found that even a slight downmodulation of CARD9 protein levels to 72% of control ( Figure 6A) led to a significant reduction in cell viability (p = 0.0018) ( Figure 6B). We also performed MALT1 substrate cylindromatosis (CYLD) Western blot analysis after CARD9 siRNA knockdown, and detected a significant reduction in C-terminal CYLD cleavage product compared with the control group (p = 0.017). Loss of CYLD cleavage products indicates reduction in MALT1 paracaspase activity [33]. Discussion We demonstrated the expression of CARD9 in the RS model system U-RT1, as well as in the parent lymphoma of the cell line. CARD9 is known to be an important activator of the NF-кB pathway in myeloid cells [25,27,34,35]. So far, there have been no published data that documented the expression or functional significance of CARD9 in non-neoplastic lymphoid cells. In the context of neoplasia, CARD9 has been demonstrated in gastric MALT lymphoma. The findings of Nakamura et al. [36] indicated a potential role of CARD9 overexpression in lymphoma progression and the development of resistance to Helicobacter pylori eradication therapy. In addition, gains of chromosomal material, including the chromosome 9q34 gene locus of CARD9, have been identified in a subset of gastric MALT lymphomas [37]. In our study, we demonstrated for the first time the expression of CARD9 in an aggressive B-cell lymphoma. One possible explanation for the high protein levels of CARD9 in U-RT1 is gene amplification, as already described in MALT-lymphoma [36,37]. Indeed, array CGH analysis revealed copy number gains involving the CARD9 gene locus in U-RT1. An alternative mechanism to be considered is enhanced gene expression by epigenetic regulation. In myeloid cells of the innate immune system, the NF-кB pathway is activated upon ligation of pathogen-associated molecular patterns (PAMPs) or endogenous danger-associated molecular patterns (DAMPs) [38,39] and subsequent stimulation through spleen tyrosine kinase (SYK)-coupled C-type lectin receptors (CLRs) [25,40,41]. In contrast to lymphoid cells, the CBM complex of myeloid cells includes CARD9 instead of CARD11 [25,27,35]. Even though CARD9 and CARD11 are highly homologous in the N-terminal CARD domain and the coiled-coiled domain for protein oligomerization and activation of the CBM signalosome, CARD9 lacks the membrane-associated guanylate kinase (MAGUK) domain, which harbors phosphorylation sites for PKCβ and PKCθ isoforms [34,42,43], and instead is activated by phosphorylation within the coiled-coiled domain at residue T231 by the PKC∆ isoform. CARD9-T231 phosphorylation induces conformational changes that disrupt the autoinhibitory state of CARD9 by exposing its CARD domain for binding of BCL10 [44,45], assembly of the CBM complex, and subsequent NF-кB activation. The kinase SYK, which is a common activator of the NF-κB pathway in lymphoid and myeloid cells, was shown to be a linker between CLRs and PKC∆ in myeloid cells [25,34,45]. Interestingly, activation of PKC∆ by SYK was found to be an antiapoptotic driver in chronic lymphocytic leukemia [46,47], and the active protein kinase SYK was also observed in follicular lymphoma [48], DLBCL [49], and mantle cell lymphoma [50]. Together, these findings indicated that PKC∆, activated by SYK, may be able to phosphorylate CARD9 in RS as well. Using co-immunoprecipitation, we demonstrated the assembly of a CBM complex with CARD9 instead of CARD11. Western blot analyses of phospho-IKK and phospho-IкB, as well as pharmacological IKK inhibition studies, indicated that the survival of U-RT1 cells depends on NF-кB activation. Importantly, knockdown of CARD9 significantly reduced U-RT1 cell viability and the quantity of CYLD cleavage products as an indicator of impaired MALT paracaspase activity, which is essential for NF-кB pathway activation. These results underlined the relevance of CARD9 in promoting the survival of U-RT1 cells, possibly via NF-кB activation. To complement our findings, we examined CARD9 protein levels in a series of RS tissue specimens and different lymphoma entities. In particular, we detected CARD9 expression in the lymphoma cells from 53% of the RS cases, whereas de novo DLBCL and DLBCL arising through transformation of lymphomas other than CLL only rarely exhibited CARD9. Out of 26 cHLs tested, CARD9 expression was detectable only in the HRS cells of a single case. In addition, 25 CLL cases remained negative for CARD9, even though 5 of the CLL cases later showed transformation into DLBCL. Although the size of our RS cohort was moderate, and further studies with more cases will be required, our results suggested that the expression of CARD9 is neither a feature of CLL, de novo DLBCL or cHL, nor does it represent a general event in the transformation of indolent lymphoma to aggressive lymphoma. Instead, CARD9 overexpression as a possible activator of the NF-кB pathway and driver of proliferation appears to be a phenomenon of RS. CARD9 could drive transformation from CLL to RS either autonomously by activating the NF-κB pathway or in combination with other genetic lesions. Chakraborty et al. demonstrated that the combined loss of TP53 and CDKN2A leads to proliferation and transformation of CLL cells in the context of BCR stimulation [18]. Because U-RT1 displays both of these genetic lesions, the question arises whether CARD9 expression could augment the effect of the underlying cell cycle inhibitor deficiency. Another mechanism of transformation that seems to be independent from CARD9 expression is activation of AKT via Notch1 [19]. Kohlhaas et al. showed that AKT was activated in high-risk CLL and in 52.6% of RS patients tested. Therefore, AKT activation might account for the transformation of CARD9-negative RS cases of our cohort. Other common lesions that frequently occur in RS, such as MYC and EGR2 activation or Trisomy 12, still need to be elucidated concerning the corresponding signaling pathway and the contribution to the transformational process. Until now, most RS patients were treated using a similar therapeutic regimen to that of patients with de novo DLBCL, even though RS and de novo DLBCL have been shown to be distinct disease entities. In view of the poor prognosis of RS patients despite intensive therapy, more targeted approaches to RS are needed. Recently, Leshchiner et al. [51] performed experiments using small-molecule inhibitors that directly targeted CARD9, and registered an attenuation of IKK phosphorylation after CARD9 inhibition in the context of inflammatory bowel disease. In our study, we described the aberrant expression of CARD9 with oncogenic features in the new RS model system U-RT1, as well as the expression of CARD9 in a subset of RS cases as a possible pathogenetic mechanism of RS for the first time. As a future prospect, CARD9 inhibition could serve as an approach in the development of novel targeted therapies for RS. Conclusions We have shown the expression of CARD9 in a subset of RS cases, in the RS model system U-RT1, as well as in the parent lymphoma of the cell line. We demonstrated the assembly of a CBM complex with CARD9 instead of CARD11. Pharmacological NF-κB inhibition and Western blot analyses of phospho-IKK and phospho-IкB indicated constitutive activation of NF-κB in U-RT1. Importantly, knockdown of CARD9 significantly reduced U-RT1 cell viability and quantity of CYLD cleavage products. These results underlined the relevance of CARD9 in promoting the survival of U-RT1 cells, possibly via NF-кB activation. This is the first report of an ectopic expression of CARD9 in an aggressive lymphoma. In view of the poor prognosis of RS patients despite intensive therapy, new therapeutic options for RS are needed. As a future prospect, CARD9 could serve as a target for therapeutical approaches in RS.	2022-01-23T16:56:16.181Z	2022-01-21T00:00:00.000	{ "year": 2022, "sha1": "f2c4fe89254c4ca96bb77ba4f5142d20948891c3", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2072-6694/14/3/531/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "b2e11d4a94d060f8e561f5ea5a123b48e25608ff", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [ "Medicine" ] }
245861010	pes2o/s2orc	v3-fos-license	Nutritional Potential and Bioactive Compounds in Cereus jamacaru DC: An Multifaceted Cactus From Brazilian Semiarid Cereus jamacaru DC. is an endemic cactus of the Brazilian semiarid region empirically used by humans for forage, food and medicinal purposes. In this paper, we characterize the physicochemical and phytochemical profile of the C. jamacaru cladode and fruits, outlining their nutritional implications. We evaluate the basic physicochemical characteristic of the fruits and the main classes of chemical compounds present in aqueous and ethanolic extracts from cladode, peel, and pulp of ripe and semi-ripe fruits through qualitatively and quantitatively methods. We analyze the data through descriptive statistics and variance analysis. The fruits have appropriate pH, acidity, and total soluble solid levels for the fruit processing industry and fresh consumption. We identified the presence of saponins, tannins, flavones, flavonols, and xanthones in all samples, but alkaloids and steroids were both detected in cladodes exclusively. The phenolic compound and flavonoid contents vary according to the extraction method and sample class. The cladodes and pulp of ripe fruits have the highest flavonoid levels, while the content of phenolic compounds had a high level in peels of ripe and semi-ripe fruits. The presence of these bioactive compounds implies that C. jamacaru products have relevant pharmacological interest and functionality for human (fruits) and domestic ruminant (cladodes) food. These applications can boost the agricultural-economic exploration of C. jamacaru and contribute to income generation, and improve human and animal nutrition. Introduction The Brazilian Seasonally Dry Tropical Forest, the so-called "Caatinga", covers an area of 912,529 km² in the semi-arid region of Northeastern Brazil (Silva, Leal and Tabarelli, 2017). The Caatinga supplies multiple environmental services and has potential for sustainable extractive activities of timber and non-timber forest products, including edible fruits and bioactive substances. However, a wide range of useful plants is still poorly explored, probably because their products were not characterized enough to drawn attention to the food and pharmaceutical sectors. Cactaceae is a plant family abundant in the Caatinga (Sampaio and Costa, 2011) that contains several useful species for humans. Cereus jamacaru DC., popularly known as mandacaru, is one of this useful Cactaceae specie for humans (Sales et al., 2014) because its fruits are edible and the stem (cladode) is a source of domestic ruminant food. The C. jamacaru cladodes are multi-articulate with candelabriform branches. Besides being an important food source for cattle during the dry season (Zara et al., 2012), the cladodes also have antimicrobial, vasodilatory (Messias, 2010), anti-inflammatory, and contraceptive actions (Andrade et al., 2006), as experimentally demonstrated in rodents. The C. jamacaru fruit is an oblong ellipsoid-shaped berry, its pulp is white funicular and mucilaginous with black-colored seeds (Rocha and Agra, 2002) frequently consumed in natura or as juices and jellies by the locals. Although the C. jamacaru phytochemical profile has been investigated before by several authors, including Bevilaqua et al. (2015), Santos et al. (2019), Dutra et al. (2019), and Medeiros et al. (2018), there are still other aspects to be studied, such as the phytochemical characterization of different parts of the plant using different extraction methods, as this would extend knowledge about its pharmacological and nutritional properties. In order to emphasize the importance of this species as a powerful source of bioactive compounds and highlight its chemical composition versatility and the possibility of many uses, we evaluated the basic physicochemical characteristic of the fruits and the main classes of chemical compounds present in aqueous and ethanolic extracts of the C. jamacaru cladodes and fruits. Collection and Preparation of the Samples We collected the fruit and cladode samples (Figure 1 D) in ten individuals (n = 10) from a C. jamacaru natural population located on private property (09 °01'49'' S, 36 °24'07'' W) in the municipality of Brejã o, Pernambuco state, Brazil. The site climatic classification based on Köppen-Geiger is Cs'a-type (mesothermal including hot and dry summer). The predominant vegetation is Subcaducifolic and Deciduous Tropical Forest-types (Santos et al., 2018). After collection, we transported the samples under refrigeration to the processing site. ISSN 2166-0379 2022 According to their color, we grouped the fruits in two maturation stages: ripe (predominantly red peel) and semi-ripe (reddish-green peel) ( Fig. 1 A and B). After fruits washed and sanitized with sodium hypochlorite (100 ppm), we manually separated the fractions of peel (epicarp + mesocarp) and pulp (endocarp + seeds) ( Fig. 1A and C), resulting in four classes of samples afterward: ripe peel (RP), semi-ripe peel (SP), ripe pulp (RP), and semi-ripe pulp (SP). These samples were frozen (-10 º C), lyophilized (Terroni Lyophilizer, model Series LS 3000), grounded, packed in hermetically-sealed glass pots, and, finally, stored at 25 º C. Journal of Agricultural Studies We properly cleaned the cladodes and removed their thorns manually. Then, they were horizontally sliced into 2.0 cm thick pieces and dried in a forced air circulation system (50 °C for 72 hours). The dry material was ground in a Cutting Mill (Willey®), packed in hermetically-sealed glass pots, and stored at 25 º C. Basic Physicochemical Characterization of the Fruits For the physicochemical characterization, we determined: a) pH -using the potentiometric method; b) total soluble solids -using the refractometry on Brix scale (°Brix); c) titratable acidity -using the volumetric titration method (% citric acid/100g of the sample). Such methodologies are available in AOC (1992). Obtaining Extracts The vegetal extracts were obtained through aqueous and ethanolic extractions. In aqueous extraction, we mixed 100g of the sample with 1.0 L of distilled water. Then, we constantly heated and shaken the mixture up to 90 °C, which remained at this temperature for 5 min. Right after, we placed the extract in hermetically-sealed glass pots and stored it at -10 º C. In ethanolic extraction, we mixed 100g of the sample with 1.0 L of ethanol (95%) -extracting solvent. The extraction was exhaustive, i.e., we put this sample and the extracting solvent together at 25 °C for eight days. We shook the mixture day after day, whereas we renewed the extracting solvent every two days. By using a rotary evaporator at 40 º C, we filtered and concentrated the mixture afterward (Vizcaino, 2007). The extract was stored in hermetically-sealed glass pots at -10 º C. ISSN 2166-0379 2022 Identification of the Main Classes of Chemical Compounds We identified the main classes of chemical compounds according to the methods proposed by Matos (2009) and Desoti et al. (2009). We used 1 mg/mL of extract (aqueous and ethanolic) for qualitative chemical tests. The steroids were detected by the Liebermann-Burchard reaction, while tannins by precipitation methods with iron salts, flavonoids through Shinoda and Taubouk reactions, as saponins by analysis of persistent foaming after extract agitation. In contrast, we performed chemical tests with specific reagents to check for the presence of phenols, flavones, flavonols, xanthones, catechins, anthocyanins, anthocyanidins, terpenoids, and flavanones, as described in table 1. The interpretation made was based on the characteristic visual patterns. Detection of the alkaloids chemical group was performed by thin-layer chromatography (CCD) using specific eluent and developer systems according to Cechinel Filho and Yunes (1998). Content of Total Phenolic Compounds (CFT) and Total Flavonoids (FT) We estimated CFT content through the modified Folin-Ciocalteau method by Roesler et al. (2007). We mixed 100 µL of extract (1:10 p / v) with 500 µL of the Folin-Ciocalteau reagent and 400 µL of Na2CO3 solution (7.5%). The samples were shaken and heated (50 º C) for 5 ISSN 2166-0379 2022 min, then cooled at 25 º C. The absorbance of the liquid fraction was measured at 760 nm using a UV-vis spectrophotometer (model 700 Femto® plus). A calibration curve of gallic acid from 50 to 300 μg mL -1 concentration (y = -0.1267 + 0.0064x, R² = 0.9892) was used to estimate the CFT content in the samples. We performed determinations in triplicate, and the results were expressed in milligrams of gallic acid equivalents in 100g of dry matter (mg EAG/100g MS). Journal of Agricultural Studies We estimated FT content according to the method described by Woisky and Salatino (1998). We mixed 0.5 ml of extract with 0.5 ml of AlCl 3 solution (5%). We left the mixture resting for 30 minutes at 25 º C with no light. The liquid fraction absorbance was measured at 425 nm using a UV-vis spectrophotometer (model 700 Femto® plus). A quercetin calibration curve from 50 to 300 μg mL -1 concentration (y = 0.0011x + 0.0078, R² = 0.9885) was used to estimate the FT content in the samples. We carried out determinations in triplicate, and the results were expressed in milligrams of quercetin equivalents in 100g of dry matter (mg EQ/100g MS). Data Analysis We used the descriptive statistics (median, standard deviation, minimum, first quartile, second quartile, third quartile, and maximum) in boxplot for data analysis concerning to basic physicochemical characterization of the fruits. As these data also attended to the parametric analysis assumptions, we used the Variance Analysis (ANOVA) to compare the treatments through the statistical software BioEstat 5.0. The data concerning both contents of total phenolic compounds and total flavonoids also attended to the parametric analysis assumptions and hence were analyzed through ANOVA in a double factorial scheme (5x2). Factor 1 concerns the sample class (cladode, ripe and semi-ripe peel, ripe and semi-ripe pulp), and the second factor concerns the extract type (aqueous and ethanolic). We compared the averages through the Tukey test at 5% probability level. We performed the analyses using the SISVAR 5.6 statistical program. Basic Physicochemical Characterization of the Fruits We noticed a higher acidity level (P = 0.0032) in the peel of semi-ripe fruits ( Figure 2A). However, the acidity (Fig. 2B) and pH (P = 0.01) levels in the pulp were similar between the two ripening stages (Fig. 2C and 2D). The content of soluble solids was higher (P<0.01) in the peel and pulp of ripe fruits (Fig. 2 -E and F). Content of Total Phenolic Compounds (CTP) and Total Flavonoids (TF) There was an interaction (P<0.05) among the extraction methods and sample classes for the CTP and TF contents. The CTP contents in cladode and peel of ripe and semi-ripe fruit were higher through ethanolic extraction. However, the pulp of ripe and semi-ripe fruit samples showed the highest CTP contents through aqueous extraction (Fig. 3). The CTP contents in the peel of ripe and semi-ripe fruits were higher than in other samples. Figure 3. Phenolic compound contents of the C. jamacaru cladode and fruit fractions in function of extraction methods. Averages followed by the same letter do not differ from each other through the Tukey test at 5% probability. Vertical bars represent the average standard The cladode and pulp of ripe fruits had higher TF contents through aqueous extraction, but peel of ripe and semi-ripe fruits had higher TF contents through ethanolic extraction (Fig. 4). The cladode and pulp of ripe fruits had the highest levels of TF (Fig. 4). Discussion In this study, we investigated the physicochemical characteristics and phytochemical composition in the C. jamacaru cladode and fruit fractions analyzing their pharmacological and nutritional implications accordingly. Physicochemical Attributes of the Fruit Fractions The pH (4.05 to 5.30) and titratable acidity (0.0087 to 0.1089% citric acid) ranges of C. jamacaru pulp indicate that this fraction has weak vulnerability to micro-organisms action, which could prolong its longevity after processing. While its total soluble solids content (SST) might be regarded as low (4.0 to 5.67 º BRIX), it resembles other native fruits from Brazil (4.0 to 4.5 º BRIX), such as Psidium cattleyanum S., Mauritia flexuosa Mart. and Byrsonima crassifolia L. Rich), being in concordance with the standard minimum criteria determined by the Brazilian legislation (BRASIL, 2016). Such characterization is relevant for the fruit processing industry because the greater the amount of SST, the greater the juice and pulp yield. In contrast, SST contents of C. jamacaru fruits (4.0 to 5.67 º BRIX) were lower than those verified by Almeida et al. (2009) (10.50 and 11.5 º BRIX), which might have happened due to the differences in the ripening stage, plant genotype, and local edaphoclimatic conditions. There has also been a pulp SST content increasing according to the fruit maturation stage, ISSN 2166-0379 2022 from 3.64 º BRIX in the semi-ripe stage to 5.21 º BRIX in the ripe one. Such increase was followed by an acidity decrease, from 0.0982% in the semi-ripe stage to 0.0570% in the ripe one. According to Damodaran et al. (2010), such change is due to the converting from citric and malic acids to sugars during the fruit ripening stage. Qualitative Identification of the Main Compound Classes We found several secondary metabolites in C. jamacaru, including alkaloids, terpenoids, phenols, and saponins. These compounds were also identified in the studies performed by The presence of these bioactive compounds suggests, therefore, C. jamacaru fruits have the functional potential for human food. In addition, the presence of terpenoids, saponins, phenols, flavones, flavonols, and xanthones in C. jamacaru cladodes also indicates that the product has the functional potential for domestic ruminants' food. A previous study performed by Davet (2005) has identified alkaloids and steroids as the main compounds contained in C. jamacaru cladodes, corroborating with the finding ones of this study and scientific literature for Cactaceae. More recently, Medeiros et al. (2018) have identified almost 24 phytochemicals in C. jamacaru, including tannins that have not been identified in our study. CTP and FT Yield According to the Extraction Method and Class of Samples We found that CTP and TF levels have varied according to the extraction method and class of samples. It probably happened because the extraction methods affect quantity, quality, and bioavailability of the metabolites (Miglio et al., 2008). The extracting solvent influences the release/expression of flavonoid groups (Koubaa et al., 2015) and phenolic compounds (Harbone, 1998), as they vary in polarity, acidity, hydrogen-bonding capacity of hydroxyl groups in the aromatic ring. While most phenolic compounds can be soluble in water, some compounds present solubility only in organic solvents. In fact, ethanolic solvents may be more efficient in the extraction of total phenolic compounds than water (Vizzoto and Pereira, 2011;Tomsone et al., 2012;Menezes Filho and Castro, 2019). However, the processing degree, the size of particles, the extraction duration, the temperature, and the extracting solvent concentration can all also influence the phenolic compound yields (Shahidi and Wanasundara, 1998). We also believe the higher CTP content in peels (177.55 -466.33 mg EAG/100g DM) regarding the C. jamacaru pulp (45.36 -91.66 mg EAG/100g DM) is probably due to an accumulation of these metabolites in the fruit epidermis because, according to Daiuto et al. (2014), such metabolites act as protection against ultraviolet radiation. Regarding the FT content in C. jamacaru cladode, the values obtained here (59.65 -123.94 EQ /100g DM) are similar to those described by Dutra et al. (2019) who used the hydroalcoholic extraction method. In contrast, TF contents in fruit peels (38.56 -92.5 mg EQ/100g DM) are higher than the ones observed by Araújo et al. (2008) (11.30 -14.34 mg/500g), indicating existence of intraspecific variability in the content of this secondary compound. This variability in Opuntia spp. is attributed to local abiotic factors such as drought and soil fertility (Figueroa -Cares et al., 2010), as they are capable of inducing changes in the composition of secondary compounds, including flavonoids (Pretti et al., 2018). These polyphenols have antioxidant, anti-cancer (Dutra et al., 2018), anti-inflammatory (Araújo et al., 2008Ferreira Júnior et al., 2011), and antibacterial (Davet et al., 2009) actions that are relevant to pharmacology. Conclusions The phenolic compound and flavonoid contents vary according to the extraction method and sample class. The cladodes and pulp of ripe fruits have the highest flavonoid contents, while the peels of ripe and semi-ripe fruits have higher phenolic compound contents. The presence of these bioactive compounds indicates that C. jamacaru products have great potential for human (fruits) and domestic ruminants (cladodes) foods and pharmacological use. These applications can boost the agricultural-economic exploration of C. jamacaru and contribute to income generation, and improve human and animal nutrition.	2022-01-12T16:21:34.365Z	2022-01-10T00:00:00.000	{ "year": 2022, "sha1": "36fb7e65165ecb6eb12bacb6298f3b2fb213c1fe", "oa_license": "CCBY", "oa_url": "https://www.macrothink.org/journal/index.php/jas/article/download/19006/15104", "oa_status": "HYBRID", "pdf_src": "Adhoc", "pdf_hash": "0f6fc436b02b75c83a792b7d52cb78aa83fe0925", "s2fieldsofstudy": [ "Agricultural And Food Sciences" ], "extfieldsofstudy": [] }

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.